From 43cee67f88ab08f9de66f3dab865cfc84d6c84bf Mon Sep 17 00:00:00 2001 From: Jerry Park Date: Wed, 26 May 2021 22:54:59 +0900 Subject: [PATCH 01/10] Update outdated files in the dev-1.21-ko.3 branch (1) --- .../windows/intro-windows-in-kubernetes.md | 1310 +++++++++++++---- 1 file changed, 995 insertions(+), 315 deletions(-) diff --git a/content/ko/docs/setup/production-environment/windows/intro-windows-in-kubernetes.md b/content/ko/docs/setup/production-environment/windows/intro-windows-in-kubernetes.md index 2a7826c414..eb48f3f65d 100644 --- a/content/ko/docs/setup/production-environment/windows/intro-windows-in-kubernetes.md +++ b/content/ko/docs/setup/production-environment/windows/intro-windows-in-kubernetes.md @@ -2,27 +2,57 @@ title: 쿠버네티스의 윈도우 지원 소개 content_type: concept weight: 65 + + + + + --- -윈도우 애플리케이션은 많은 조직에서 실행되는 서비스 및 애플리케이션의 상당 부분을 구성한다. [윈도우 컨테이너](https://aka.ms/windowscontainers)는 프로세스와 패키지 종속성을 캡슐화하는 현대적인 방법을 제공하여, 데브옵스(DevOps) 사례를 더욱 쉽게 ​​사용하고 윈도우 애플리케이션의 클라우드 네이티브 패턴을 따르도록 한다. 쿠버네티스는 사실상의 표준 컨테이너 오케스트레이터가 되었으며, 쿠버네티스 1.14 릴리스에는 쿠버네티스 클러스터의 윈도우 노드에서 윈도우 컨테이너 스케줄링을 위한 프로덕션 지원이 포함되어 있어, 광범위한 윈도우 애플리케이션 생태계가 쿠버네티스의 강력한 기능을 활용할 수 있다. 윈도우 기반 애플리케이션과 리눅스 기반 애플리케이션에 투자한 조직은 워크로드를 관리하기 위해 별도의 오케스트레이터를 찾을 필요가 없으므로, 운영 체제와 관계없이 배포 전반에 걸쳐 운영 효율성이 향상된다. +윈도우 애플리케이션은 많은 조직에서 실행되는 서비스 및 +애플리케이션의 상당 부분을 구성한다. +[윈도우 컨테이너](https://aka.ms/windowscontainers)는 프로세스와 패키지 종속성을 +캡슐화하는 현대적인 방법을 제공하여, 데브옵스(DevOps) +사례를 더욱 쉽게 사용하고 윈도우 애플리케이션의 클라우드 네이티브 패턴을 따르도록 한다. +쿠버네티스는 사실상의 표준 컨테이너 오케스트레이터가 되었으며, +쿠버네티스 1.14 릴리스에는 쿠버네티스 클러스터의 윈도우 노드에서 윈도우 +컨테이너 스케줄링을 위한 프로덕션 지원이 포함되어 있어, 광범위한 윈도우 애플리케이션 생태계가 +쿠버네티스의 강력한 기능을 활용할 수 있다. 윈도우 기반 애플리케이션과 +리눅스 기반 애플리케이션에 투자한 조직은 워크로드를 관리하기 위해 +별도의 오케스트레이터를 찾을 필요가 없으므로, +운영 체제와 관계없이 배포 전반에 걸쳐 +운영 효율성이 향상된다. ## 쿠버네티스의 윈도우 컨테이너 -쿠버네티스에서 윈도우 컨테이너 오케스트레이션을 활성화하려면, 기존 리눅스 클러스터에 윈도우 노드를 포함한다. 쿠버네티스의 {{< glossary_tooltip text="파드" term_id="pod" >}}에서 윈도우 컨테이너를 스케줄링하는 것은 리눅스 기반 컨테이너를 스케줄링하는 것과 유사하다. +쿠버네티스에서 윈도우 컨테이너 오케스트레이션을 활성화하려면, 기존 +리눅스 클러스터에 윈도우 노드를 포함한다. 쿠버네티스의 +{{< glossary_tooltip text="파드" term_id="pod" >}}에서 윈도우 컨테이너를 스케줄링하는 것은 +리눅스 기반 컨테이너를 스케줄링하는 것과 유사하다. -윈도우 컨테이너를 실행하려면, 쿠버네티스 클러스터에 리눅스를 실행하는 컨트롤 플레인 노드와 사용자의 워크로드 요구에 따라 윈도우 또는 리눅스를 실행하는 워커가 있는 여러 운영 체제가 포함되어 있어야 한다. 윈도우 서버 2019는 윈도우에서 [쿠버네티스 노드](https://github.com/kubernetes/community/blob/master/contributors/design-proposals/architecture/architecture.md#the-kubernetes-node)를 활성화하는 유일한 윈도우 운영 체제이다(kubelet, [컨테이너 런타임](https://docs.microsoft.com/ko-kr/virtualization/windowscontainers/deploy-containers/containerd) 및 kube-proxy 포함). 윈도우 배포 채널에 대한 자세한 설명은 [Microsoft 문서](https://docs.microsoft.com/ko-kr/windows-server/get-started-19/servicing-channels-19)를 참고한다. +윈도우 컨테이너를 실행하려면, 쿠버네티스 클러스터에 리눅스를 +실행하는 컨트롤 플레인 노드와 사용자의 워크로드 요구에 따라 윈도우 또는 리눅스를 +실행하는 워커가 있는 여러 운영 체제가 포함되어 있어야 한다. 윈도우 +서버 2019는 윈도우에서 +[쿠버네티스 노드](https://github.com/kubernetes/community/blob/master/contributors/design-proposals/architecture/architecture.md#the-kubernetes-node)를 +활성화하는 유일한 윈도우 운영 체제이다(kubelet, +[컨테이너 런타임](https://docs.microsoft.com/ko-kr/virtualization/windowscontainers/deploy-containers/containerd) +및 kube-proxy 포함). 윈도우 배포 채널에 대한 자세한 설명은 +[Microsoft 문서](https://docs.microsoft.com/ko-kr/windows-server/get-started-19/servicing-channels-19)를 참고한다. -{{< note >}} -[마스터 컴포넌트](/ko/docs/concepts/overview/components/)를 포함한 쿠버네티스 컨트롤 플레인은 리눅스에서 계속 실행된다. 윈도우 전용 쿠버네티스 클러스터는 계획이 없다. -{{< /note >}} +[마스터 컴포넌트](/ko/docs/concepts/overview/components/)를 포함한 +쿠버네티스 컨트롤 플레인은 +리눅스에서 계속 실행된다. +윈도우 전용 쿠버네티스 클러스터는 계획이 없다. -{{< note >}} -이 문서에서 윈도우 컨테이너에 대해 이야기할 때 프로세스 격리된 윈도우 컨테이너를 의미한다. [Hyper-V 격리](https://docs.microsoft.com/ko-kr/virtualization/windowscontainers/manage-containers/hyperv-container)가 있는 윈도우 컨테이너는 향후 릴리스로 계획되어 있다. -{{< /note >}} +이 문서에서 윈도우 컨테이너에 대해 이야기할 때 +프로세스 격리된 윈도우 컨테이너를 의미한다. +[Hyper-V 격리](https://docs.microsoft.com/ko-kr/virtualization/windowscontainers/manage-containers/hyperv-container)가 +있는 윈도우 컨테이너는 향후 릴리스로 계획되어 있다. ## 지원되는 기능 및 제한 @@ -30,41 +60,68 @@ weight: 65 #### 윈도우 OS 버전 지원 -쿠버네티스의 윈도우 운영 체제 지원은 다음 표를 참조한다. 단일 이기종 쿠버네티스 클러스터에는 윈도우 및 리눅스 워커 노드가 모두 있을 수 있다. 윈도우 컨테이너는 윈도우 노드에서, 리눅스 컨테이너는 리눅스 노드에서 스케줄되어야 한다. +쿠버네티스의 윈도우 운영 체제 지원은 다음 표를 +참조한다. 단일 이기종 쿠버네티스 클러스터에는 윈도우 및 +리눅스 워커 노드가 모두 있을 수 있다. 윈도우 컨테이너는 윈도우 노드에서, +리눅스 컨테이너는 리눅스 노드에서 스케줄되어야 한다. | 쿠버네티스 버전 | 윈도우 서버 LTSC 릴리스 | 윈도우 서버 SAC 릴리스 | -| --- | --- | --- | -| *Kubernetes v1.17* | Windows Server 2019 | Windows Server ver 1809 | -| *Kubernetes v1.18* | Windows Server 2019 | Windows Server ver 1809, Windows Server ver 1903, Windows Server ver 1909 | +| --- | --- | --- | --- | | *Kubernetes v1.19* | Windows Server 2019 | Windows Server ver 1909, Windows Server ver 2004 | | *Kubernetes v1.20* | Windows Server 2019 | Windows Server ver 1909, Windows Server ver 2004 | +| *Kubernetes v1.21* | Windows Server 2019 | Windows Server ver 2004, Windows Server ver 20H2 | -{{< note >}} -지원 모델을 포함한 다양한 윈도우 서버 서비스 채널에 대한 정보는 [윈도우 서버 서비스 채널](https://docs.microsoft.com/ko-kr/windows-server/get-started-19/servicing-channels-19)에서 확인할 수 있다. -{{< /note >}} -{{< note >}} -모든 윈도우 고객이 앱의 운영 체제를 자주 업데이트하는 것은 아니다. 애플리케이션 업그레이드를 위해서는 클러스터에 새 노드를 업그레이드하거나 도입하는 것이 필요하다. 이 문서에서 쿠버네티스에서 실행되는 컨테이너의 운영 체제를 업그레이드하기로 선택한 고객을 위해 새 운영 체제 버전에 대한 지원을 추가할 때의 가이드와 단계별 지침을 제공한다. 이 가이드에는 클러스터 노드와 함께 사용자 애플리케이션을 업그레이드하기 위한 권장 업그레이드 절차가 포함된다. 윈도우 노드는 현재 리눅스 노드와 동일한 방식으로 쿠버네티스 [버전-스큐(skew) 정책](/ko/docs/setup/release/version-skew-policy/)(노드 대 컨트롤 플레인 버전 관리)을 준수한다. -{{< /note >}} -{{< note >}} -윈도우 서버 호스트 운영 체제에는 [윈도우 서버](https://www.microsoft.com/ko-kr/cloud-platform/windows-server-pricing) 라이선스가 적용된다. 윈도우 컨테이너 이미지에는 [윈도우 컨테이너에 대한 추가 사용 조건](https://docs.microsoft.com/en-us/virtualization/windowscontainers/images-eula)이 적용된다. -{{< /note >}} -{{< note >}} -프로세스 격리가 포함된 윈도우 컨테이너에는 엄격한 호환성 규칙이 있으며, [여기서 호스트 OS 버전은 컨테이너 베이스 이미지 OS 버전과 일치해야 한다](https://docs.microsoft.com/ko-kr/virtualization/windowscontainers/deploy-containers/version-compatibility). 일단 쿠버네티스에서 Hyper-V 격리가 포함된 윈도우 컨테이너를 지원하면, 제한 및 호환성 규칙이 변경될 것이다. -{{< /note >}} + +지원 모델을 포함한 다양한 윈도우 서버 +서비스 채널에 대한 정보는 +[윈도우 서버 서비스 채널](https://docs.microsoft.com/ko-kr/windows-server/get-started-19/servicing-channels-19)에서 확인할 수 있다. + +모든 윈도우 고객이 앱의 운영 체제를 자주 업데이트하는 것은 +아니다. 애플리케이션 업그레이드를 위해서는 클러스터에 새 노드를 +업그레이드하거나 도입하는 것이 필요하다. 이 문서에서 +쿠버네티스에서 실행되는 컨테이너의 운영 체제를 업그레이드하기로 선택한 +고객을 위해 새 운영 체제 버전에 대한 지원을 추가할 때의 가이드와 +단계별 지침을 제공한다. 이 가이드에는 클러스터 노드와 함께 사용자 애플리케이션을 +업그레이드하기 위한 권장 업그레이드 절차가 포함된다. +윈도우 노드는 현재 리눅스 노드와 동일한 방식으로 쿠버네티스 +[버전-스큐(skew) 정책](/ko/docs/setup/release/version-skew-policy/)(노드 대 컨트롤 플레인 +버전 관리)을 준수한다. + + +윈도우 서버 호스트 운영 체제에는 +[윈도우 서버](https://www.microsoft.com/ko-kr/cloud-platform/windows-server-pricing) +라이선스가 적용된다. 윈도우 컨테이너 이미지에는 +[윈도우 컨테이너에 대한 추가 사용 조건](https://docs.microsoft.com/en-us/virtualization/windowscontainers/images-eula)이 적용된다. + +프로세스 격리가 포함된 윈도우 컨테이너에는 엄격한 호환성 규칙이 있으며, +[여기서 호스트 OS 버전은 컨테이너 베이스 이미지 OS 버전과 일치해야 한다](https://docs.microsoft.com/ko-kr/virtualization/windowscontainers/deploy-containers/version-compatibility). +일단 쿠버네티스에서 Hyper-V 격리가 포함된 윈도우 컨테이너를 지원하면, +제한 및 호환성 규칙이 변경될 것이다. #### 퍼즈(Pause) 이미지 -Microsoft는 `mcr.microsoft.com/oss/kubernetes/pause:1.4.1`에서 윈도우 퍼즈 인프라 컨테이너를 유지한다. +Microsoft는 `mcr.microsoft.com/oss/kubernetes/pause:3.4.1`에서 +윈도우 퍼즈 인프라 컨테이너를 유지한다. #### 컴퓨트 -API 및 kubectl의 관점에서, 윈도우 컨테이너는 리눅스 기반 컨테이너와 거의 같은 방식으로 작동한다. 그러나 [제한 섹션](#제한)에 요약된 주요 기능에는 몇 가지 눈에 띄는 차이점이 있다. +API 및 kubectl의 관점에서, 윈도우 컨테이너는 +리눅스 기반 컨테이너와 거의 같은 방식으로 작동한다. 그러나 +[제한 섹션](#제한)에 요약된 주요 기능에는 +몇 가지 눈에 띄는 차이점이 있다. -윈도우에서 주요 쿠버네티스 요소는 리눅스와 동일한 방식으로 작동한다. 이 섹션에서는, 주요 워크로드 인에이블러(enabler) 일부와 이들이 윈도우에 매핑되는 방법에 대해 설명한다. +윈도우에서 주요 쿠버네티스 요소는 리눅스와 동일한 방식으로 작동한다. 이 +섹션에서는, 주요 워크로드 인에이블러(enabler) 일부와 이들이 윈도우에 매핑되는 방법에 +대해 설명한다. * [파드](/ko/docs/concepts/workloads/pods/) - 파드는 쿠버네티스의 기본 빌딩 블록이다 - 쿠버네티스 오브젝트 모델에서 생성하고 배포하는 가장 작고 간단한 단위. 동일한 파드에 윈도우 및 리눅스 컨테이너를 배포할 수 없다. 파드의 모든 컨테이너는 단일 노드로 스케줄되며 각 노드는 특정 플랫폼 및 아키텍처를 나타낸다. 다음과 같은 파드 기능, 속성 및 이벤트가 윈도우 컨테이너에서 지원된다. + 파드는 쿠버네티스의 기본 빌딩 블록이다 - 쿠버네티스 오브젝트 모델에서 + 생성하고 배포하는 가장 작고 간단한 단위. 동일한 파드에 + 윈도우 및 리눅스 컨테이너를 배포할 수 없다. 파드의 모든 컨테이너는 + 단일 노드로 스케줄되며 각 노드는 특정 플랫폼 및 + 아키텍처를 나타낸다. 다음과 같은 파드 기능, 속성 및 + 이벤트가 윈도우 컨테이너에서 지원된다. * 프로세스 분리 및 볼륨 공유 기능을 갖춘 파드 당 하나 또는 여러 개의 컨테이너 * 파드 상태 필드 @@ -76,7 +133,8 @@ API 및 kubectl의 관점에서, 윈도우 컨테이너는 리눅스 기반 컨 * 리소스 제한 * [컨트롤러](/ko/docs/concepts/workloads/controllers/) - 쿠버네티스 컨트롤러는 파드의 의도한 상태(desired state)를 처리한다. 윈도우 컨테이너에서 지원되는 워크로드 컨트롤러는 다음과 같다. + 쿠버네티스 컨트롤러는 파드의 의도한 상태(desired state)를 처리한다. 윈도우 + 컨테이너에서 지원되는 워크로드 컨트롤러는 다음과 같다. * 레플리카셋(ReplicaSet) * 레플리케이션컨트롤러(ReplicationController) @@ -87,7 +145,10 @@ API 및 kubectl의 관점에서, 윈도우 컨테이너는 리눅스 기반 컨 * 크론잡(CronJob) * [서비스](/ko/docs/concepts/services-networking/service/) - 쿠버네티스 서비스는 논리적인 파드 집합과 그것에(마이크로 서비스라고도 함) 접근하는 정책을 정의하는 추상화 개념이다. 상호-운영 체제 연결을 위해 서비스를 사용할 수 있다. 윈도우에서 서비스는 다음의 유형, 속성 및 기능을 활용할 수 있다. + 쿠버네티스 서비스는 논리적인 파드 집합과 그것에(마이크로 서비스라고도 함) + 접근하는 정책을 정의하는 추상화 개념이다. 상호-운영 체제 + 연결을 위해 서비스를 사용할 수 있다. 윈도우에서 서비스는 + 다음의 유형, 속성 및 기능을 활용할 수 있다. * 서비스 환경 변수 * 노드포트(NodePort) @@ -96,7 +157,10 @@ API 및 kubectl의 관점에서, 윈도우 컨테이너는 리눅스 기반 컨 * ExternalName * 헤드리스 서비스(Headless services) -파드, 컨트롤러 및 서비스는 쿠버네티스에서 윈도우 워크로드를 관리하는데 중요한 요소이다. 그러나 그 자체로는 동적 클라우드 네이티브 환경에서 윈도우 워크로드의 적절한 수명 주기 관리를 수행하기에 충분하지 않다. 다음 기능에 대한 지원이 추가되었다. +파드, 컨트롤러 및 서비스는 쿠버네티스에서 윈도우 워크로드를 +관리하는데 중요한 요소이다. 그러나 그 자체로는 동적 클라우드 네이티브 환경에서 +윈도우 워크로드의 적절한 수명 주기 관리를 수행하기에 +충분하지 않다. 다음 기능에 대한 지원이 추가되었다. * 파드와 컨테이너 메트릭 * Horizontal Pod Autoscaler 지원 @@ -110,27 +174,42 @@ API 및 kubectl의 관점에서, 윈도우 컨테이너는 리눅스 기반 컨 {{< feature-state for_k8s_version="v1.14" state="stable" >}} -Docker EE-basic 19.03 이상은 모든 윈도우 서버 버전에 대해 권장되는 컨테이너 런타임이다. 이것은 kubelet에 포함된 dockershim 코드와 함께 작동한다. +Docker EE-basic 19.03 이상은 모든 윈도우 서버 버전에 대해 권장되는 +컨테이너 런타임이다. 이것은 kubelet에 포함된 dockershim 코드와 함께 작동한다. ##### CRI-ContainerD {{< feature-state for_k8s_version="v1.20" state="stable" >}} -{{< glossary_tooltip term_id="containerd" text="ContainerD" >}} 1.4.0+는 윈도우 쿠버네티스 노드의 컨테이너 런타임으로도 사용할 수 있다. +{{< glossary_tooltip term_id="containerd" text="ContainerD" >}} 1.4.0+는 +윈도우 쿠버네티스 노드의 컨테이너 런타임으로도 사용할 수 있다. -[윈도우에 ContainerD 설치](/ko/docs/setup/production-environment/container-runtimes/#containerd-설치) 방법을 확인한다. - -{{< caution >}} -ContainerD와 함께 GMSA를 사용하여 커널 패치가 필요한 윈도우 네트워크 공유에 액세스 할 때 [알려진 제한](/docs/tasks/configure-pod-container/configure-gmsa/#gmsa-limitations)이 있다. 이 제한을 해결하기위한 업데이트는 현재 Windows Server, 버전 2004에서 사용할 수 있으며 2021년 초에 Windows Server 2019에서 사용할 수 있다. [Microsoft 윈도우 컨테이너 이슈 트래커](https://github.com/microsoft/Windows-Containers/issues/44)에서 업데이트를 확인한다. -{{< /caution >}} +[윈도우에 ContainerD 설치](/ko/docs/setup/production-environment/container-runtimes/#containerd-설치) +방법을 확인한다. #### 퍼시스턴트 스토리지(Persistent Storage) -쿠버네티스 [볼륨](/ko/docs/concepts/storage/volumes/)을 사용하면 데이터 지속성(persistence) 및 파드 볼륨 공유 요구 사항이 있는 복잡한 애플리케이션을 쿠버네티스에 배포할 수 있다. 특정 스토리지 백엔드 또는 프로토콜과 관련된 퍼시스턴트 볼륨 관리에는 볼륨 프로비저닝/디-프로비저닝/크기 조정, 쿠버네티스 노드에 볼륨 연결/분리, 데이터를 유지해야 하는 파드의 개별 컨테이너에 볼륨 마운트/분리와 같은 작업이 포함된다. 특정 스토리지 백엔드 또는 프로토콜에 대해 이러한 볼륨 관리 작업을 구현하는 코드는 쿠버네티스 볼륨 [플러그인](/ko/docs/concepts/storage/volumes/#볼륨-유형들)의 형태로 제공된다. 다음과 같은 광범위한 쿠버네티스 볼륨 플러그인 클래스가 윈도우에서 지원된다. +쿠버네티스 [볼륨](/ko/docs/concepts/storage/volumes/)을 사용하면 +데이터 지속성(persistence) 및 파드 볼륨 공유 요구 사항이 있는 복잡한 애플리케이션을 +쿠버네티스에 배포할 수 있다. 특정 스토리지 백엔드 또는 +프로토콜과 관련된 퍼시스턴트 볼륨 관리에는 +볼륨 프로비저닝/디-프로비저닝/크기 조정, 쿠버네티스 노드에 볼륨 +연결/분리, 데이터를 유지해야 하는 파드의 개별 컨테이너에 볼륨 +마운트/분리와 같은 작업이 포함된다. 특정 스토리지 백엔드 또는 +프로토콜에 대해 이러한 볼륨 관리 작업을 +구현하는 코드는 쿠버네티스 볼륨 +[플러그인](/ko/docs/concepts/storage/volumes/#볼륨-유형들)의 형태로 제공된다. 다음과 같은 +광범위한 쿠버네티스 볼륨 플러그인 클래스가 윈도우에서 지원된다. ##### 인-트리(In-tree) 볼륨 플러그인 -인-트리 볼륨 플러그인과 관련된 코드는 핵심 쿠버네티스 코드 베이스의 일부로 제공된다. 인-트리 볼륨 플러그인 배포는 추가 스크립트를 설치하거나 별도의 컨테이너화된 플러그인 컴포넌트를 배포할 필요가 없다. 이러한 플러그인들은 볼륨 프로비저닝/디-프로비저닝, 스토리지 백엔드 볼륨 크기 조정, 쿠버네티스 노드에 볼륨 연결/분리, 파드의 개별 컨테이너에 볼륨 마운트/분리를 처리할 수 있다. 다음의 인-트리 플러그인은 윈도우 노드를 지원한다. +인-트리 볼륨 플러그인과 관련된 코드는 핵심 쿠버네티스 +코드 베이스의 일부로 제공된다. 인-트리 볼륨 플러그인 배포는 +추가 스크립트를 설치하거나 별도의 컨테이너화된 플러그인 컴포넌트를 +배포할 필요가 없다. 이러한 플러그인들은 +볼륨 프로비저닝/디-프로비저닝, 스토리지 백엔드 볼륨 크기 조정, 쿠버네티스 노드에 +볼륨 연결/분리, 파드의 개별 컨테이너에 볼륨 마운트/분리를 +처리할 수 있다. 다음의 인-트리 플러그인은 윈도우 노드를 지원한다. * [awsElasticBlockStore](/ko/docs/concepts/storage/volumes/#awselasticblockstore) * [azureDisk](/ko/docs/concepts/storage/volumes/#azuredisk) @@ -140,7 +219,16 @@ ContainerD와 함께 GMSA를 사용하여 커널 패치가 필요한 윈도우 ##### FlexVolume 플러그인 -[FlexVolume](/ko/docs/concepts/storage/volumes/#flexVolume) 플러그인과 관련된 코드는 아웃-오브-트리(out-of-tree) 스크립트 또는 호스트에 직접 배포해야 하는 바이너리로 제공된다. FlexVolume 플러그인은 쿠버네티스 노드에 볼륨 연결/분리 및 파드의 개별 컨테이너에 볼륨 마운트/분리를 처리한다. FlexVolume 플러그인과 관련된 퍼시스턴트 볼륨의 프로비저닝/디-프로비저닝은 일반적으로 FlexVolume 플러그인과는 별도의 외부 프로비저너를 통해 처리될 수 있다. 호스트에서 powershell 스크립트로 배포된 다음의 FlexVolume [플러그인](https://github.com/Microsoft/K8s-Storage-Plugins/tree/master/flexvolume/windows)은 윈도우 노드를 지원한다. +[FlexVolume](/ko/docs/concepts/storage/volumes/#flexVolume) +플러그인과 관련된 코드는 아웃-오브-트리(out-of-tree) 스크립트 또는 호스트에 직접 배포해야 하는 +바이너리로 제공된다. FlexVolume 플러그인은 쿠버네티스 노드에 볼륨 +연결/분리 및 파드의 개별 컨테이너에 볼륨 마운트/분리를 +처리한다. FlexVolume 플러그인과 관련된 퍼시스턴트 볼륨의 +프로비저닝/디-프로비저닝은 일반적으로 FlexVolume 플러그인과는 별도의 외부 +프로비저너를 통해 처리될 수 있다. 호스트에서 +powershell 스크립트로 배포된 다음의 FlexVolume +[플러그인](https://github.com/Microsoft/K8s-Storage-Plugins/tree/master/flexvolume/windows)은 +윈도우 노드를 지원한다. * [SMB](https://github.com/microsoft/K8s-Storage-Plugins/tree/master/flexvolume/windows/plugins/microsoft.com~smb.cmd) * [iSCSI](https://github.com/microsoft/K8s-Storage-Plugins/tree/master/flexvolume/windows/plugins/microsoft.com~iscsi.cmd) @@ -149,13 +237,40 @@ ContainerD와 함께 GMSA를 사용하여 커널 패치가 필요한 윈도우 {{< feature-state for_k8s_version="v1.19" state="beta" >}} -{{< glossary_tooltip text="CSI" term_id="csi" >}} 플러그인과 관련된 코드는 일반적으로 컨테이너 이미지로 배포되고 데몬셋(DaemonSets) 및 스테이트풀셋(StatefulSets)과 같은 표준 쿠버네티스 구성을 사용하여 배포되는 아웃-오브-트리 스크립트 및 바이너리로 제공된다. CSI 플러그인은 쿠버네티스에서 볼륨 프로비저닝/디-프로비저닝, 볼륨 크기 조정, 쿠버네티스 노드에 볼륨 연결/분리, 파드의 개별 컨테이너에 볼륨 마운트/분리, 스냅샷 및 복제를 사용하여 퍼시스턴트 데이터 백업/복원과 같은 다양한 볼륨 관리 작업을 처리한다. CSI 플러그인은 일반적으로 (각 노드에서 데몬셋으로 실행되는) 노드 플러그인과 컨트롤러 플러그인으로 구성된다. +{{< glossary_tooltip text="CSI" term_id="csi" >}} 플러그인과 +관련된 코드는 일반적으로 컨테이너 이미지로 배포되고 데몬셋(DaemonSets) +및 스테이트풀셋(StatefulSets)과 같은 +표준 쿠버네티스 구성을 사용하여 배포되는 아웃-오브-트리 스크립트 및 +바이너리로 제공된다. CSI 플러그인은 쿠버네티스에서 볼륨 프로비저닝/디-프로비저닝, 볼륨 +크기 조정, 쿠버네티스 노드에 볼륨 연결/분리, 파드의 개별 컨테이너에 볼륨 +마운트/분리, 스냅샷 및 복제를 사용하여 퍼시스턴트 데이터 백업/복원과 같은 +다양한 볼륨 관리 작업을 처리한다. CSI 플러그인은 +일반적으로 (각 노드에서 데몬셋으로 실행되는) 노드 플러그인과 컨트롤러 +플러그인으로 구성된다. -CSI 노드 플러그인(특히 블록 디바이스 또는 공유 파일시스템으로 노출된 퍼시스턴트 볼륨과 관련된 플러그인)은 디스크 장치 스캔, 파일 시스템 마운트 등과 같은 다양한 특권이 필요한(privileged) 작업을 수행해야 한다. 이러한 작업은 호스트 운영 체제마다 다르다. 리눅스 워커 노드의 경우 컨테이너화된 CSI 노드 플러그인은 일반적으로 특권을 가진 컨테이너로 배포된다. 윈도우 워커 노드의 경우 컨테이너화된 CSI 노드 플러그인에 대한 특권이 필요한 작업은 커뮤니티에서 관리되고, 각 윈도우 노드에 사전 설치되어야 하는 독립형(stand-alone) 바이너리인 [csi-proxy](https://github.com/kubernetes-csi/csi-proxy)를 사용하여 지원된다. 자세한 내용은 배포하려는 CSI 플러그인의 배포 가이드를 참조한다. +CSI 노드 플러그인(특히 블록 디바이스 또는 공유 파일시스템으로 노출된 +퍼시스턴트 볼륨과 관련된 플러그인)은 디스크 장치 스캔, 파일 시스템 마운트 등과 같은 +다양한 특권이 필요한(privileged) 작업을 수행해야 +한다. 이러한 작업은 호스트 운영 체제마다 다르다. 리눅스 워커 +노드의 경우 컨테이너화된 CSI 노드 플러그인은 일반적으로 특권을 가진 +컨테이너로 배포된다. 윈도우 워커 노드의 경우 컨테이너화된 +CSI 노드 플러그인에 대한 특권이 필요한 작업은 커뮤니티에서 관리되고, +각 윈도우 노드에 사전 설치되어야 하는 독립형(stand-alone) 바이너리인 +[csi-proxy](https://github.com/kubernetes-csi/csi-proxy)를 사용하여 지원된다. 자세한 +내용은 배포하려는 CSI 플러그인의 배포 가이드를 +참조한다. #### 네트워킹 -윈도우 컨테이너용 네트워킹은 [CNI 플러그인](/ko/docs/concepts/extend-kubernetes/compute-storage-net/network-plugins/)을 통해 노출된다. 윈도우 컨테이너는 네트워킹과 관련하여 가상 머신과 유사하게 작동한다. 각 컨테이너에는 Hyper-V 가상 스위치(vSwitch)에 연결된 가상 네트워크 어댑터(vNIC)가 있다. 호스트 네트워킹 서비스(HNS)와 호스트 컴퓨팅 서비스(HCS)는 함께 작동하여 컨테이너를 만들고 컨테이너 vNIC을 네트워크에 연결한다. HCS는 컨테이너 관리를 담당하는 반면 HNS는 다음과 같은 네트워킹 리소스 관리를 담당한다. +윈도우 컨테이너용 네트워킹은 +[CNI 플러그인](/ko/docs/concepts/extend-kubernetes/compute-storage-net/network-plugins/)을 통해 노출된다. +윈도우 컨테이너는 네트워킹과 관련하여 가상 머신과 유사하게 +작동한다. 각 컨테이너에는 Hyper-V 가상 스위치(vSwitch)에 연결된 +가상 네트워크 어댑터(vNIC)가 있다. 호스트 네트워킹 서비스(HNS)와 +호스트 컴퓨팅 서비스(HCS)는 함께 작동하여 컨테이너를 만들고 +컨테이너 vNIC을 네트워크에 연결한다. HCS는 컨테이너 관리를 +담당하는 반면 HNS는 다음과 같은 네트워킹 리소스 관리를 +담당한다. * 가상 네트워크(vSwitch 생성 포함) * 엔드포인트 / vNIC @@ -171,19 +286,155 @@ CSI 노드 플러그인(특히 블록 디바이스 또는 공유 파일시스템 ##### 네트워크 모드 -윈도우는 L2bridge, L2tunnel, Overlay, Transparent 및 NAT의 다섯 가지 네트워킹 드라이버/모드를 지원한다. 윈도우와 리눅스 워커 노드가 있는 이기종 클러스터에서는 윈도우와 리눅스 모두에서 호환되는 네트워킹 솔루션을 선택해야 한다. 윈도우에서 다음과 같은 out-of-tree 플러그인이 지원되며 각 CNI 사용 시 권장 사항이 있다. +윈도우는 L2bridge, L2tunnel, Overlay, Transparent 및 +NAT의 다섯 가지 네트워킹 드라이버/모드를 지원한다. 윈도우와 리눅스 워커 노드가 +있는 이기종 클러스터에서는 윈도우와 리눅스 모두에서 호환되는 네트워킹 +솔루션을 선택해야 한다. 윈도우에서 다음과 같은 out-of-tree 플러그인이 지원되며 +각 CNI 사용 시 권장 사항이 있다. -| 네트워크 드라이버 | 설명 | 컨테이너 패킷 수정 | 네트워크 플러그인 | 네트워크 플러그인 특성 | -| -------------- | ----------- | ------------------------------ | --------------- | ------------------------------ | -| L2bridge | 컨테이너는 외부 vSwitch에 연결된다. 컨테이너는 언더레이 네트워크에 연결된다. 하지만 인그레스/이그레스시에 재작성되기 때문에 물리적 네트워크가 컨테이너 MAC을 학습할 필요가 없다. | MAC은 호스트 MAC에 다시 쓰여지고, IP는 HNS OutboundNAT 정책을 사용하여 호스트 IP에 다시 쓰여질 수 있다. | [win-bridge](https://github.com/containernetworking/plugins/tree/master/plugins/main/windows/win-bridge), [Azure-CNI](https://github.com/Azure/azure-container-networking/blob/master/docs/cni.md), Flannel 호스트 게이트웨이는 win-bridge를 사용한다. | win-bridge는 L2bridge 네트워크 모드를 사용하고, 컨테이너를 호스트의 언더레이에 연결하여 최상의 성능을 제공한다. 노드 간 연결을 위해 사용자 정의 경로(user-defined routes, UDR)가 필요하다. | -| L2Tunnel | 이것은 l2bridge의 특별한 케이스이지만 Azure에서만 사용된다. 모든 패킷은 SDN 정책이 적용되는 가상화 호스트로 전송된다. | MAC 재작성되고, 언더레이 네트워크 상에서 IP가 보인다. | [Azure-CNI](https://github.com/Azure/azure-container-networking/blob/master/docs/cni.md) | Azure-CNI를 사용하면 컨테이너를 Azure vNET과 통합할 수 있으며, [Azure Virtual Network에서 제공하는](https://azure.microsoft.com/ko-kr/services/virtual-network/) 기능 집합을 활용할 수 있다. 예를 들어 Azure 서비스에 안전하게 연결하거나 Azure NSG를 사용한다. [azure-cni 예제](https://docs.microsoft.com/ko-kr/azure/aks/concepts-network#azure-cni-advanced-networking)를 참고한다. | -| 오버레이(쿠버네티스에서 윈도우용 오버레이 네트워킹은 *알파* 단계에 있음) | 컨테이너에는 외부 vSwitch에 연결된 vNIC이 제공된다. 각 오버레이 네트워크는 사용자 지정 IP 접두사로 정의된 자체 IP 서브넷을 가져온다. 오버레이 네트워크 드라이버는 VXLAN 캡슐화를 사용한다. | 외부 헤더로 캡슐화된다. | [Win-overlay](https://github.com/containernetworking/plugins/tree/master/plugins/main/windows/win-overlay), Flannel VXLAN(win-overlay 사용) | win-overlay는 가상 컨테이너 네트워크를 호스트의 언더레이에서 격리하려는 경우(예: 보안 상의 이유로) 사용해야 한다. 데이터 센터의 IP에 제한이 있는 경우, (다른 VNID 태그가 있는) 다른 오버레이 네트워크에 IP를 재사용할 수 있다. 이 옵션을 사용하려면 윈도우 서버 2019에서 [KB4489899](https://support.microsoft.com/help/4489899)가 필요하다. | -| Transparent([ovn-kubernetes](https://github.com/openvswitch/ovn-kubernetes)의 특수한 유스케이스) | 외부 vSwitch가 필요하다. 컨테이너는 논리적 네트워크(논리적 스위치 및 라우터)를 통해 파드 내 통신을 가능하게 하는 외부 vSwitch에 연결된다. | 패킷은 [GENEVE](https://datatracker.ietf.org/doc/draft-gross-geneve/) 또는 [STT](https://datatracker.ietf.org/doc/draft-davie-stt)를 통해 캡슐화되는데, 동일한 호스트에 있지 않은 파드에 도달하기 위한 터널링을 한다.
패킷은 ovn 네트워크 컨트롤러에서 제공하는 터널 메타데이터 정보를 통해 전달되거나 삭제된다.
NAT는 north-south 통신(데이터 센터와 클라이언트, 네트워크 상의 데이터 센터 외부와의 통신)을 위해 수행된다. | [ovn-kubernetes](https://github.com/openvswitch/ovn-kubernetes) | [ansible을 통해 배포](https://github.com/openvswitch/ovn-kubernetes/tree/master/contrib)한다. 분산 ACL은 쿠버네티스 정책을 통해 적용할 수 있다. IPAM을 지원한다. kube-proxy 없이 로드 밸런싱을 수행할 수 있다. NAT를 수행할 때 iptables/netsh를 사용하지 않고 수행된다. | -| NAT(*쿠버네티스에서 사용되지 않음*) | 컨테이너에는 내부 vSwitch에 연결된 vNIC이 제공된다. DNS/DHCP는 [WinNAT](https://blogs.technet.microsoft.com/virtualization/2016/05/25/windows-nat-winnat-capabilities-and-limitations/)라는 내부 컴포넌트를 사용하여 제공된다. | MAC 및 IP는 호스트 MAC/IP에 다시 작성된다. | [nat](https://github.com/Microsoft/windows-container-networking/tree/master/plugins/nat) | 완전성을 위해 여기에 포함되었다. | + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +
네트워크 드라이버설명컨테이너 패킷 수정네트워크 플러그인네트워크 플러그인 특성
L2bridge컨테이너는 외부 vSwitch에 연결된다. 컨테이너는 + 언더레이 네트워크에 연결된다. 하지만 인그레스/이그레스시에 재작성되기 + 때문에 물리적 네트워크가 컨테이너 MAC을 학습할 필요가 없다. + + MAC은 호스트 MAC에 다시 쓰여지고, IP는 HNS OutboundNAT 정책을 사용하여 + 호스트 IP에 다시 쓰여질 수 있다. + + win-bridge, + Azure-CNI, + Flannel 호스트 게이트웨이는 win-bridge를 사용한다. + + win-bridge는 L2bridge 네트워크 모드를 사용하고, + 컨테이너를 호스트의 언더레이에 연결하여 최상의 성능을 제공한다. + 노드 간 연결을 위해 사용자 정의 경로(user-defined routes, UDR)가 필요하다. +
L2Tunnel + 이것은 l2bridge의 특별한 케이스이지만 Azure에서만 사용된다. 모든 패킷은 + SDN 정책이 적용되는 가상화 호스트로 전송된다. + + MAC 재작성되고, 언더레이 네트워크 상에서 IP가 보인다. + + Azure-CNI + + Azure-CNI를 사용하면 컨테이너를 Azure vNET과 통합할 수 있으며, + Azure Virtual Network에서 + 제공하는 기능 집합을 활용할 수 있다. + 예를 들어, Azure 서비스에 안전하게 연결하거나 Azure NSG를 사용한다. + azure-cni + 예제를 참고한다. +
오버레이(쿠버네티스에서 윈도우용 오버레이 네트워킹은 알파 단계에 있음) + 컨테이너에는 외부 vSwitch에 연결된 vNIC이 제공된다. 각 오버레이 + 네트워크는 사용자 지정 IP 접두사로 정의된 자체 IP 서브넷을 가져온다. 오버레이 + 네트워크 드라이버는 VXLAN 캡슐화를 사용한다. + + 외부 헤더로 캡슐화된다. + + Win-overlay, + Flannel VXLAN (win-overlay 사용) + + win-overlay는 가상 컨테이너 네트워크를 호스트의 + 언더레이에서 격리하려는 경우(예: 보안 상의 이유로) 사용해야 한다. 데이터 센터의 IP에 + 제한이 있는 경우, (다른 VNID 태그가 있는) 다른 오버레이 + 네트워크에 IP를 재사용할 수 있다. 이 옵션을 사용하려면 + 윈도우 서버 2019에서 KB4489899가 + 필요하다. +
+ Transparent(ovn-kubernetes의 특수한 유스케이스) + + 외부 vSwitch가 필요하다. 컨테이너는 논리적 네트워크(논리적 스위치 및 라우터)를 + 통해 파드 내 통신을 가능하게 하는 외부 vSwitch에 + 연결된다. + + 패킷은 + GENEVE, + STT 터널링을 통해 + 캡슐화되는데, 동일한 호스트에 있지 않은 파드에 도달하기 위한 터널링을 한다.
패킷은 ovn 네트워크 + 컨트롤러에서 제공하는 터널 메타데이터 정보를 통해 전달되거나 삭제된다. +
+ NAT는 north-south 통신(데이터 센터와 클라이언트, 네트워크 상의 데이터 센터 외부와의 통신)을 위해 수행된다. + 		+ ovn-kubernetes + + Ansible을 통해 배포한다. + 분산 ACL은 쿠버네티스 정책을 통해 적용할 수 있다. IPAM을 지원한다. + kube-proxy 없이 로드 밸런싱을 수행할 수 있다. NAT를 수행할 때 + iptables/netsh를 사용하지 않고 수행된다. +
NAT (쿠버네티스에서 사용되지 않음) + 컨테이너에는 내부 vSwitch에 연결된 vNIC이 제공된다. DNS/DHCP는 + WinNAT라는 + 내부 컴포넌트를 사용하여 제공된다. + + MAC 및 IP는 호스트 MAC/IP에 다시 작성된다. + + nat + + 완전성을 위해 여기에 포함되었다. +
-위에서 설명한대로 [플란넬(Flannel)](https://github.com/coreos/flannel) CNI [메타 플러그인](https://github.com/containernetworking/plugins/tree/master/plugins/meta/flannel)은 [VXLAN 네트워크 백엔드](https://github.com/coreos/flannel/blob/master/Documentation/backends.md#vxlan)(**alpha 지원**, win-overlay에 위임) 및 [host-gateway network backend](https://github.com/coreos/flannel/blob/master/Documentation/backends.md#host-gw) (안정적인 지원, win-bridge에 위임)를 통해 [윈도우](https://github.com/containernetworking/plugins/tree/master/plugins/meta/flannel#windows-support-experimental)에서도 지원된다. 이 플러그인은 자동 노드 서브넷 임대 할당과 HNS 네트워크 생성을 위해 윈도우 (Flanneld)에서 Flannel 데몬과 함께 작동하도록 참조 CNI 플러그인 (win-overlay, win-bridge) 중 하나에 대한 위임을 지원한다. 이 플러그인은 자체 구성 파일 (cni.conf)을 읽고, 이를 FlannelD 생성하는 subnet.env 파일의 환경 변수와 함께 집계한다. 이후 네트워크 연결을 위한 참조 CNI 플러그인 중 하나에 위임하고 노드 할당 서브넷을 포함하는 올바른 구성을 IPAM 플러그인 (예: 호스트-로컬)으로 보낸다. +위에서 설명한대로 [플란넬(Flannel)](https://github.com/coreos/flannel) CNI +[메타 플러그인](https://github.com/containernetworking/plugins/tree/master/plugins/meta/flannel)은 +[VXLAN 네트워크 백엔드](https://github.com/coreos/flannel/blob/master/Documentation/backends.md#vxlan) +(**alpha 지원**, win-overlay에 위임) 및 +[host-gateway network backend](https://github.com/coreos/flannel/blob/master/Documentation/backends.md#host-gw) +(안정적인 지원, win-bridge에 위임)를 통해 +[윈도우](https://github.com/containernetworking/plugins/tree/master/plugins/meta/flannel#windows-support-experimental)에서도 +지원된다. 이 플러그인은 자동 노드 서브넷 +임대 할당과 HNS 네트워크 생성을 위해 윈도우 (Flanneld)에서 +Flannel 데몬과 함께 작동하도록 참조 CNI 플러그인 (win-overlay, win-bridge) +중 하나에 대한 위임을 지원한다. 이 플러그인은 자체 +구성 파일 (cni.conf)을 읽고, 이를 FlannelD 생성하는 subnet.env 파일의 환경 변수와 +함께 집계한다. 이후 네트워크 연결을 위한 +참조 CNI 플러그인 중 하나에 위임하고 노드 할당 서브넷을 포함하는 올바른 +구성을 IPAM 플러그인 (예: 호스트-로컬)으로 +보낸다. -노드, 파드, 서비스 오브젝트의 경우 TCP/UDP 트래픽에 대해 다음 네트워크 흐름이 지원된다. +노드, 파드, 서비스 오브젝트의 경우 TCP/UDP 트래픽에 대해 다음 +네트워크 흐름이 지원된다. * 파드 -> 파드(IP) * 파드 -> 파드(Name) @@ -205,84 +456,227 @@ CSI 노드 플러그인(특히 블록 디바이스 또는 공유 파일시스템 ##### 로드 밸런싱과 서비스 -윈도우에서는 다음 설정을 사용하여 서비스 및 로드 밸런싱 동작을 구성할 수 있다. +윈도우에서는 다음 설정을 사용하여 서비스 및 로드 밸런싱 동작을 +구성할 수 있다. {{< table caption="윈도우 서비스 구성" >}} -| 기능 | 설명 | 지원되는 쿠버네티스 버전 | 지원되는 윈도우 OS 빌드 | 활성화하는 방법 | -| ------- | ----------- | ----------------------------- | -------------------------- | ------------- | -| 세션 어피니티 | 특정 클라이언트의 연결이 매번 동일한 파드로 전달되도록 한다. | v1.20 이상 | [윈도우 서버 vNext Insider Preview Build 19551](https://blogs.windows.com/windowsexperience/2020/01/28/announcing-windows-server-vnext-insider-preview-build-19551/) 이상 | `service.spec.sessionAffinity`를 "ClientIP"로 설정 | -| 직접 서버 반환 (DSR) | IP 주소 수정 및 LBNAT가 컨테이너 vSwitch 포트에서 직접 발생하는 로드 밸런싱 모드. 서비스 트래픽은 소스 IP가 원래 파드 IP로 설정된 상태로 도착한다. | v1.20 이상 | 윈도우 서버 2019 | kube-proxy에서 다음 플래그를 설정한다. `--feature-gates="WinDSR=true" --enable-dsr=true` | -| 대상 보존(Preserve-Destination) | 서비스 트래픽의 DNAT를 스킵하여, 백엔드 파드에 도달하는 패킷에서 대상 서비스의 가상 IP를 보존한다. 또한 노드-노드 전달을 비활성화한다. | v1.20 이상 | 윈도우 서버, 버전 1903 (또는 그 이상) | 서비스 어노테이션에서 `"preserve-destination": "true"`를 설정하고 kube-proxy에서 DSR을 활성화한다. | -| IPv4/IPv6 이중 스택 네트워킹 | 클러스터 내/외부 기본 IPv4-to-IPv4 통신과 함께 IPv6-to-IPv6 통신 | v1.19 이상 | 윈도우 서버, 버전 2004 (또는 그 이상) | [IPv4/IPv6 이중 스택](#ipv4ipv6-이중-스택)을 참고한다. | -| 클라이언트 IP 보존 | 인그레스 트래픽의 소스 IP가 유지되도록 한다. 또한 노드-노드 전달을 비활성화한다. | v1.20 이상 | 윈도우 서버, 버전 2019 (또는 그 이상) | `service.spec.externalTrafficPolicy` 를 "Local"로 설정하고 kube-proxy에서 DSR을 활성화한다. | + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +
기능설명지원되는 쿠버네티스 버전지원되는 윈도우 OS 빌드활성화하는 방법
세션 어피니티 + 특정 클라이언트의 연결이 매번 동일한 파드로 + 전달되도록 한다. + v1.20 이상 + 윈도우 서버 vNext Insider Preview Build 19551 (또는 그 이상) + + service.spec.sessionAffinity를 "ClientIP"로 설정 +
직접 서버 반환 (DSR) + IP 주소 수정 및 LBNAT가 컨테이너 vSwitch 포트에서 직접 + 발생하는 로드 밸런싱 모드. 서비스 트래픽은 소스 IP가 원래 파드 IP로 + 설정된 상태로 도착한다. + v1.20 이상 + 윈도우 서버 2019 + + kube-proxy에서 다음 플래그를 설정한다. + --feature-gates="WinDSR=true" --enable-dsr=true +
대상 보존(Preserve-Destination) + 서비스 트래픽의 DNAT를 스킵하여, 백엔드 파드에 도달하는 패킷에서 대상 + 서비스의 가상 IP를 보존한다. 또한 노드-노드 전달을 비활성화한다. + v1.20 이상윈도우 서버, 버전 1903 (또는 그 이상) + 서비스 어노테이션에서 "preserve-destination": "true"를 설정하고 + kube-proxy에서 DSR을 활성화한다. +
IPv4/IPv6 이중 스택 네트워킹 + 클러스터 내/외부 기본 IPv4-to-IPv4 통신과 함께 + IPv6-to-IPv6 통신 + v1.19 이상윈도우 서버, 버전 2004 (또는 그 이상) + IPv4/IPv6 이중 스택을 참고한다. +
클라이언트 IP 보존 + 인그레스 트래픽의 소스 IP가 유지되도록 한다. 또한 + 노드-노드 전달을 비활성화한다. + v1.20 이상윈도우 서버, 버전 2019 (또는 그 이상) + service.spec.externalTrafficPolicy를 "Local"로 설정하고 + kube-proxy에서 DSR을 활성화한다. +
+ {{< /table >}} #### IPv4/IPv6 이중 스택 -`IPv6DualStack` [기능 게이트](/ko/docs/reference/command-line-tools-reference/feature-gates/)를 사용하여 `l2bridge` 네트워크에 IPv4/IPv6 이중 스택 네트워킹을 활성화할 수 있다. 자세한 내용은 [IPv4/IPv6 이중 스택 활성화](/ko/docs/concepts/services-networking/dual-stack/#ipv4-ipv6-이중-스택-활성화)를 참조한다. +`IPv6DualStack` [기능 게이트](/ko/docs/reference/command-line-tools-reference/feature-gates/)를 +사용하여 `l2bridge` 네트워크에 IPv4/IPv6 이중 스택 네트워킹을 활성화할 수 있다. 자세한 내용은 +[IPv4/IPv6 이중 스택 활성화](/ko/docs/concepts/services-networking/dual-stack/#ipv4-ipv6-이중-스택-활성화)를 +참조한다. -{{< note >}} -윈도우에서 쿠버네티스와 함께 IPv6를 사용하려면 윈도우 서버 버전 2004 (커널 버전 10.0.19041.610) 이상이 필요하다. -{{< /note >}} +윈도우에서 쿠버네티스와 함께 IPv6를 사용하려면 윈도우 서버 버전 2004 +(커널 버전 10.0.19041.610) 이상이 필요하다. -{{< note >}} 윈도우의 오버레이(VXLAN) 네트워크는 현재 이중 스택 네트워킹을 지원하지 않는다. -{{< /note >}} ### 제한 -윈도우는 쿠버네티스 아키텍처 및 컴포넌트 매트릭스에서 워커 노드로만 지원된다. 즉, 쿠버네티스 클러스터에는 항상 리눅스 마스터 노드가 반드시 포함되어야 하고, 0개 이상의 리눅스 워커 노드 및 0개 이상의 윈도우 워커 노드가 포함된다. +윈도우는 쿠버네티스 아키텍처 및 컴포넌트 매트릭스에서 워커 +노드로만 지원된다. 즉, 쿠버네티스 클러스터에는 항상 리눅스 마스터 노드가 반드시 +포함되어야 하고, 0개 이상의 리눅스 워커 노드 및 0개 이상의 윈도우 +워커 노드가 포함된다. #### 자원 관리 - 리눅스 cgroup은 리눅스에서 리소스 제어를 위한 파드 경계로 사용된다. 컨테이너는 네트워크, 프로세스 및 파일시스템 격리를 위해 해당 경계 내에 생성된다. cgroups API는 cpu/io/memory 통계를 수집하는 데 사용할 수 있다. 반대로 윈도우는 시스템 네임스페이스 필터가 있는 컨테이너별로 잡(Job) 오브젝트를 사용하여 컨테이너의 모든 프로세스를 포함하고 호스트와의 논리적 격리를 제공한다. 네임스페이스 필터링 없이 윈도우 컨테이너를 실행할 수 있는 방법은 없다. 즉, 시스템 권한은 호스트 컨텍스트에서 삽입 될(assert) 수 없으므로 권한이 있는(privileged) 컨테이너는 윈도우에서 사용할 수 없다. 보안 계정 매니져(Security Account Manager, SAM)가 분리되어 있으므로 컨테이너는 호스트의 ID를 가정할 수 없다. +리눅스 cgroup은 리눅스에서 리소스 제어를 위한 파드 경계로 사용된다. +컨테이너는 네트워크, 프로세스 및 파일시스템 격리를 위해 해당 +경계 내에 생성된다. cgroups API는 cpu/io/memory 통계를 수집하는 데 사용할 수 있다. +반대로 윈도우는 시스템 네임스페이스 필터가 있는 컨테이너별로 잡(Job) +오브젝트를 사용하여 컨테이너의 모든 프로세스를 포함하고 호스트와의 +논리적 격리를 제공한다. 네임스페이스 필터링 없이 윈도우 컨테이너를 +실행할 수 있는 방법은 없다. 즉, 시스템 권한은 호스트 컨텍스트에서 삽입될(assert) 수 없으므로 +권한이 있는(privileged) 컨테이너는 윈도우에서 사용할 수 없다. 보안 계정 +매니져(Security Account Manager, SAM)가 분리되어 있으므로 +컨테이너는 호스트의 ID를 가정할 수 없다. #### 자원 예약 ##### 메모리 예약 -윈도우에는 리눅스에는 있는 메모리 부족 프로세스 킬러가 없다. 윈도우는 모든 사용자-모드 메모리 할당을 항상 가상 메모리처럼 처리하며, 페이지파일이 필수이다. 결과적으로 윈도우에서는 리눅스에서 발생할 수 있는 메모리 부족 상태에 도달하지 않으며, 프로세스는 메모리 부족 (out of memory, OOM) 종료를 겪는 대신 디스크로 페이징한다. 메모리가 오버프로비저닝되고 모든 물리 메모리가 고갈되면 페이징으로 인해 성능이 저하될 수 있다. -kubelet 파라미터 `--kubelet-reserve` 를 사용하여 메모리 사용량을 합리적인 범위 내로 유지할 수 있으며, `--system-reserve` 를 사용하여 노드 (컨테이너 외부) 의 메모리 사용량을 예약할 수 있다. 이들을 사용하면 그만큼 [노드 할당(NodeAllocatable)](/docs/tasks/administer-cluster/reserve-compute-resources/#node-allocatable)은 줄어든다. +윈도우에는 리눅스에는 있는 메모리 부족 프로세스 킬러가 없다. 윈도우는 +모든 사용자-모드 메모리 할당을 항상 가상 메모리처럼 처리하며, 페이지파일이 +필수이다. 결과적으로 윈도우에서는 리눅스에서 발생할 수 있는 +메모리 부족 상태에 도달하지 않으며, 프로세스는 메모리 부족(out of memory, OOM) 종료를 +겪는 대신 디스크로 페이징한다. 메모리가 오버프로비저닝되고 +모든 물리 메모리가 고갈되면 페이징으로 인해 성능이 저하될 수 있다. -{{< note >}} -워크로드를 배포할 때, 컨테이너에 리소스 제한을 걸어라 (제한만 설정하거나, 제한이 요청과 같아야 함). 이 또한 NodeAllocatable 에서 차감되며, 메모리가 꽉 찬 노드에 스케줄러가 파드를 할당하지 않도록 제한한다. -{{< /note >}} +kubelet 파라미터 `--kubelet-reserve` 를 사용하여 메모리 사용량을 +합리적인 범위 내로 유지할 수 있으며, `--system-reserve` 를 사용하여 +노드(컨테이너 외부)의 메모리 사용량을 예약할 수 있다. 이들을 사용하면 그만큼 +[노드 할당(NodeAllocatable)](/docs/tasks/administer-cluster/reserve-compute-resources/#node-allocatable)은 줄어든다. -오버프로비저닝을 방지하는 가장 좋은 방법은 윈도우, 도커, 그리고 쿠버네티스 프로세스를 위해 최소 2GB 이상의 시스템 예약 메모리로 kubelet을 설정하는 것이다. +워크로드를 배포할 때, 컨테이너에 리소스 제한을 +걸어라(제한만 설정하거나, 제한이 요청과 같아야 함). 이 또한 NodeAllocatable에서 차감되며, +메모리가 꽉 찬 노드에 스케줄러가 파드를 할당하지 않도록 제한한다. + +오버프로비저닝을 방지하는 가장 좋은 방법은 윈도우, 도커, 그리고 +쿠버네티스 프로세스를 위해 최소 2GB 이상의 시스템 예약 메모리로 +kubelet을 설정하는 것이다. ##### CPU 예약 -윈도우, 도커, 그리고 다른 쿠버네티스 호스트 프로세스가 이벤트에 잘 응답할 수 있도록, CPU의 일정 비율을 예약하는 것이 좋다. 이 값은 윈도우 노드에 있는 CPU 코어 수에 따라 조정해야 한다. 이 비율을 결정하려면, 각 노드의 최대 파드 밀도(density)를 관찰하고, 시스템 서비스의 CPU 사용량을 모니터링하여 워크로드 요구사항을 충족하는 값을 선택해야 한다. -kubelet 파라미터 `--kubelet-reserve` 를 사용하여 CPU 사용량을 합리적인 범위 내로 유지할 수 있으며, `--system-reserve` 를 사용하여 노드 (컨테이너 외부) 의 CPU 사용량을 예약할 수 있다. 이들을 사용하면 그만큼 [노드 할당(NodeAllocatable)](/docs/tasks/administer-cluster/reserve-compute-resources/#node-allocatable)은 줄어든다. +윈도우, 도커, 그리고 다른 쿠버네티스 호스트 프로세스가 이벤트에 +잘 응답할 수 있도록, CPU의 일정 비율을 예약하는 것이 +좋다. 이 값은 윈도우 노드에 있는 CPU 코어 수에 +따라 조정해야 한다. 이 비율을 결정하려면, 각 노드의 +최대 파드 밀도(density)를 관찰하고, 시스템 서비스의 CPU +사용량을 모니터링하여 워크로드 요구사항을 충족하는 값을 선택해야 한다. + +kubelet 파라미터 `--kubelet-reserve` 를 사용하여 CPU 사용량을 +합리적인 범위 내로 유지할 수 있으며, `--system-reserve` 를 사용하여 +노드 (컨테이너 외부) 의 CPU 사용량을 예약할 수 있다. 이들을 사용하면 그만큼 +[노드 할당(NodeAllocatable)](/docs/tasks/administer-cluster/reserve-compute-resources/#node-allocatable)은 줄어든다. #### 기능 제한 + * TerminationGracePeriod: 구현되지 않음 * 단일 파일 매핑: CRI-ContainerD로 구현 예정 * 종료 메시지: CRI-ContainerD로 구현 예정 * 특권을 가진(Privileged) 컨테이너: 현재 윈도우 컨테이너에서 지원되지 않음 * HugePages: 현재 윈도우 컨테이너에서 지원되지 않음 -* 기존 노드 문제 감지기는 리눅스 전용이며 특권을 가진 컨테이너가 필요하다. 윈도우에서 특권을 가진 컨테이너를 지원하지 않기 때문에 일반적으로 윈도우에서 이 기능이 사용될 것으로 예상하지 않는다. -* 공유 네임스페이스의 모든 기능이 지원되는 것은 아니다. (자세한 내용은 API 섹션 참조). +* 기존 노드 문제 감지기는 리눅스 전용이며 특권을 가진 + 컨테이너가 필요하다. 윈도우에서 특권을 가진 컨테이너를 지원하지 않기 때문에 + 일반적으로 윈도우에서 이 기능이 사용될 것으로 예상하지 않는다. +* 공유 네임스페이스의 모든 기능이 지원되는 것은 아니다. (자세한 내용은 + API 섹션 참조). #### 각 플래그의 리눅스와의 차이점 + 윈도우 노드에서의 kubelet 플래그의 동작은 아래에 설명된 대로 다르게 동작한다. -* `--kubelet-reserve`, `--system-reserve`, `--eviction-hard` 플래그는 Node Allocatable 업데이트 +* `--kubelet-reserve`, `--system-reserve`, `--eviction-hard` 플래그는 + Node Allocatable 업데이트 + * `--enforce-node-allocable`을 사용한 축출(Eviction)은 구현되지 않았다. + * `--eviction-hard`와 `--eviction-soft`를 사용한 축출은 구현되지 않았다. + * MemoryPressure 조건은 구현되지 않았다. + * kubelet이 취한 OOM 축출 조치가 없다. -* 윈도우 노드에서 실행되는 Kubelet에는 메모리 제한이 없다. `--kubelet-reserve`와 `--system-reserve`는 호스트에서 실행되는 kubelet 또는 프로세스에 제한을 설정하지 않는다. 이는 호스트의 kubelet 또는 프로세스가 node-allocatable 및 스케줄러 외부에서 메모리 리소스 부족을 유발할 수 있음을 의미한다. -* kubelet 프로세스의 우선 순위를 설정하는 추가 플래그는 `--windows-priorityclass`라는 윈도우 노드에서 사용할 수 있다. 이 플래그를 사용하면 kubelet 프로세스가 윈도우 호스트에서 실행중인 다른 프로세스와 비교할 때 더 많은 CPU 시간 슬라이스을 얻을 수 있다. 허용되는 값과 그 의미에 대한 자세한 내용은 [윈도우 우선순위 클래스](https://docs.microsoft.com/en-us/windows/win32/procthread/scheduling-priorities#priority-class)에서 확인할 수 있다. kubelet이 항상 충분한 CPU주기를 갖도록 하려면 이 플래그를 `ABOVE_NORMAL_PRIORITY_CLASS` 이상으로 설정하는 것이 좋다. + +* 윈도우 노드에서 실행되는 Kubelet에는 메모리 제한이 없다. + `--kubelet-reserve`와 `--system-reserve`는 호스트에서 실행되는 kubelet 또는 + 프로세스에 제한을 설정하지 않는다. 이는 호스트의 kubelet 또는 프로세스가 + node-allocatable 및 스케줄러 외부에서 메모리 리소스 부족을 유발할 수 있음을 + 의미한다. + +* kubelet 프로세스의 우선 순위를 설정하는 추가 플래그는 + `--windows-priorityclass`라는 윈도우 노드에서 사용할 수 있다. 이 플래그를 사용하면 + kubelet 프로세스가 윈도우 호스트에서 실행중인 다른 프로세스와 비교할 때 더 많은 CPU 시간 + 슬라이스을 얻을 수 있다. 허용되는 값과 그 의미에 대한 자세한 내용은 + [윈도우 우선순위 클래스](https://docs.microsoft.com/en-us/windows/win32/procthread/scheduling-priorities#priority-class)에서 + 확인할 수 있다. + kubelet이 항상 충분한 CPU주기를 갖도록 하려면 + 이 플래그를 `ABOVE_NORMAL_PRIORITY_CLASS` 이상으로 설정하는 것이 좋다. #### 스토리지 -윈도우에는 컨테이너 계층을 마운트하고 NTFS를 기반으로 하는 복제 파일시스템을 만드는 레이어드(layered) 파일시스템 드라이버가 있다. 컨테이너의 모든 파일 경로는 해당 컨테이너의 컨텍스트 내에서만 확인된다. +윈도우에는 컨테이너 계층을 마운트하고 NTFS를 기반으로 하는 복제 파일시스템을 +만드는 레이어드(layered) 파일시스템 드라이버가 있다. 컨테이너의 모든 파일 경로는 +해당 컨테이너의 컨텍스트 내에서만 확인된다. -* 도커 볼륨 마운트는 개별 파일이 아닌 컨테이너의 디렉토리 만 대상으로 할 수 있다. 이 제한은 CRI-containerD에는 존재하지 않는다. -* 볼륨 마운트는 파일이나 디렉터리를 호스트 파일시스템으로 다시 투영할 수 없다. -* 읽기 전용 파일시스템은 윈도우 레지스트리 및 SAM 데이터베이스에 항상 쓰기 접근이 필요하기 때문에 지원되지 않는다. 그러나 읽기 전용 볼륨은 지원된다. -* 볼륨 사용자 마스크(user-masks) 및 권한은 사용할 수 없다. SAM은 호스트와 컨테이너 간에 공유되지 않기 때문에 이들 간에 매핑이 없다. 모든 권한은 컨테이너 컨텍스트 내에서 해결된다. +* 도커 볼륨 마운트는 개별 파일이 아닌 컨테이너의 + 디렉터리만 대상으로 할 수 있다. 이 제한은 CRI-containerD에는 존재하지 않는다. + +* 볼륨 마운트는 파일이나 디렉터리를 호스트 파일시스템으로 다시 + 투영할 수 없다. + +* 읽기 전용 파일시스템은 윈도우 레지스트리 및 SAM 데이터베이스에 항상 + 쓰기 접근이 필요하기 때문에 지원되지 않는다. 그러나 읽기 전용 + 볼륨은 지원된다. + +* 볼륨 사용자 마스크(user-masks) 및 권한은 사용할 수 없다. SAM은 + 호스트와 컨테이너 간에 공유되지 않기 때문에 이들 간에 매핑이 없다. 모든 + 권한은 컨테이너 컨텍스트 내에서 해결된다. 결과적으로, 다음 스토리지 기능은 윈도우 노드에서 지원되지 않는다. @@ -299,24 +693,61 @@ kubelet 파라미터 `--kubelet-reserve` 를 사용하여 CPU 사용량을 합 #### 네트워킹 {#네트워킹-제한} -윈도우 컨테이너 네트워킹은 리눅스 네트워킹과 몇 가지 중요한 면에서 다르다. [윈도우 컨테이너 네트워킹에 대한 Microsoft 문서](https://docs.microsoft.com/ko-kr/virtualization/windowscontainers/container-networking/architecture)에는 추가 세부 정보와 배경이 포함되어 있다. +윈도우 컨테이너 네트워킹은 리눅스 네트워킹과 몇 가지 중요한 면에서 +다르다. [윈도우 컨테이너 네트워킹에 대한 Microsoft 문서](https://docs.microsoft.com/ko-kr/virtualization/windowscontainers/container-networking/architecture)에는 +추가 세부 정보와 배경이 포함되어 있다. -윈도우 호스트 네트워킹 서비스와 가상 스위치는 네임스페이스를 구현하고 파드 또는 컨테이너에 필요한 가상 NIC을 만들 수 있다. 그러나 DNS, 라우트, 메트릭과 같은 많은 구성은 리눅스에서와 같이 /etc/... 파일이 아닌 윈도우 레지스트리 데이터베이스에 저장된다. 컨테이너의 윈도우 레지스트리는 호스트 레지스트리와 별개이므로 호스트에서 컨테이너로 /etc/resolv.conf를 매핑하는 것과 같은 개념은 리눅스에서와 동일한 효과를 갖지 않는다. 해당 컨테이너의 컨텍스트에서 실행되는 윈도우 API를 사용하여 구성해야 한다. 따라서 CNI 구현에서는 파일 매핑에 의존하는 대신 HNS를 호출하여 네트워크 세부 정보를 파드 또는 컨테이너로 전달해야 한다. +윈도우 호스트 네트워킹 서비스와 가상 스위치는 네임스페이스를 +구현하고 파드 또는 컨테이너에 필요한 가상 NIC을 만들 수 있다. 그러나 +DNS, 라우트, 메트릭과 같은 많은 구성은 리눅스에서와 같이 /etc/... 파일이 +아닌 윈도우 레지스트리 데이터베이스에 저장된다. 컨테이너의 +윈도우 레지스트리는 호스트 레지스트리와 별개이므로 호스트에서 +컨테이너로 /etc/resolv.conf를 매핑하는 것과 같은 개념은 리눅스에서와 +동일한 효과를 갖지 않는다. 해당 컨테이너의 컨텍스트에서 실행되는 윈도우 API를 +사용하여 구성해야 한다. 따라서 CNI 구현에서는 파일 매핑에 의존하는 +대신 HNS를 호출하여 네트워크 세부 정보를 파드 또는 컨테이너로 +전달해야 한다. 다음 네트워킹 기능은 윈도우 노드에서 지원되지 않는다. * 윈도우 파드에서는 호스트 네트워킹 모드를 사용할 수 없다. -* 노드 자체에서 로컬 NodePort 접근은 실패한다. (다른 노드 또는 외부 클라이언트에서는 가능) -* 노드에서 서비스 VIP에 접근하는 것은 향후 윈도우 서버 릴리스에서 사용할 수 있다. + +* 노드 자체에서 로컬 NodePort 접근은 실패한다. (다른 노드 또는 + 외부 클라이언트에서는 가능) + +* 노드에서 서비스 VIP에 접근하는 것은 향후 윈도우 서버 릴리스에서 + 사용할 수 있다. + * 한 서비스는 최대 64개의 백엔드 파드 또는 고유한 목적지 IP를 지원할 수 있다. -* kube-proxy의 오버레이 네트워킹 지원은 베타 기능이다. 또한 윈도우 서버 2019에 [KB4482887](https://support.microsoft.com/ko-kr/help/4482887/windows-10-update-kb4482887)을 설치해야 한다. + +* kube-proxy의 오버레이 네트워킹 지원은 베타 기능이다. 또한 + 윈도우 서버 2019에 [KB4482887](https://support.microsoft.com/ko-kr/help/4482887/windows-10-update-kb4482887)을 + 설치해야 한다. + * 비-DSR 모드의 로컬 트래픽 정책 -* 오버레이 네트워크에 연결된 윈도우 컨테이너는 IPv6 스택을 통한 통신을 지원하지 않는다. 이 네트워크 드라이버가 IPv6 주소를 사용하고 kubelet, kube-proxy 및 CNI 플러그인에서 후속 쿠버네티스 작업을 사용할 수 있도록 하는데 필요한 뛰어난 윈도우 플랫폼 작업이 있다. -* win-overlay, win-bridge, Azure-CNI 플러그인을 통해 ICMP 프로토콜을 사용하는 아웃바운드 통신. 특히, 윈도우 데이터 플레인([VFP](https://www.microsoft.com/en-us/research/project/azure-virtual-filtering-platform/))은 ICMP 패킷 치환을 지원하지 않는다. 이것은 다음을 의미한다. - * 동일한 네트워크(예: ping을 통한 파드 간 통신) 내의 목적지로 전달되는 ICMP 패킷은 예상대로 제한 없이 작동한다. + +* 오버레이 네트워크에 연결된 윈도우 컨테이너는 + IPv6 스택을 통한 통신을 지원하지 않는다. 이 네트워크 드라이버가 IPv6 주소를 + 사용하고 kubelet, kube-proxy 및 CNI 플러그인에서 후속 쿠버네티스 작업을 + 사용할 수 있도록 하는데 필요한 뛰어난 윈도우 플랫폼 작업이 있다. + +* win-overlay, win-bridge, Azure-CNI 플러그인을 통해 + ICMP 프로토콜을 사용하는 아웃바운드 통신. 특히, 윈도우 데이터 플레인 + ([VFP](https://www.microsoft.com/en-us/research/project/azure-virtual-filtering-platform/))은 + ICMP 패킷 치환을 지원하지 않는다. 이것은 다음을 의미한다. + + * 동일한 네트워크(예: ping을 통한 파드 간 통신) 내의 목적지로 전달되는 + ICMP 패킷은 예상대로 제한 없이 작동한다. + * TCP/UDP 패킷은 예상대로 제한 없이 작동한다. - * 원격 네트워크를 통과하도록 지정된 ICMP 패킷(예: ping을 통한 파드에서 외부 인터넷으로의 통신)은 치환될 수 없으므로 소스로 다시 라우팅되지 않는다. - * TCP/UDP 패킷은 여전히 ​​치환될 수 있기 때문에 `ping `을 `curl `으로 대체하여 외부와의 연결을 디버깅할 수 있다. + + * 원격 네트워크를 통과하도록 지정된 ICMP 패킷(예: ping을 통한 + 파드에서 외부 인터넷으로의 통신)은 치환될 수 없으므로 + 소스로 다시 라우팅되지 않는다. + + * TCP/UDP 패킷은 여전히 ​​치환될 수 있기 때문에 + `ping `을 `curl `으로 대체하여 + 외부와의 연결을 디버깅할 수 있다. 해당 기능은 쿠버네티스 v1.15에 추가되었다. @@ -324,334 +755,583 @@ kubelet 파라미터 `--kubelet-reserve` 를 사용하여 CPU 사용량을 합 ##### CNI 플러그인 -* 윈도우 참조 네트워크 플러그인 win-bridge와 win-overlay는 현재 "CHECK" 구현 누락으로 인해 [CNI 사양](https://github.com/containernetworking/cni/blob/master/SPEC.md) v0.4.0을 구현하지 않는다. +* 윈도우 참조 네트워크 플러그인 win-bridge와 win-overlay는 + 현재 "CHECK" 구현 누락으로 인해 [CNI 사양](https://github.com/containernetworking/cni/blob/master/SPEC.md) + v0.4.0을 구현하지 않는다. + * Flannel VXLAN CNI는 윈도우에서 다음과 같은 제한이 있다. -1. 노드-파드 연결은 설계상 불가능하다. Flannel v0.12.0(또는 그 이상)이 있는 로컬 파드에서만 가능하다. -2. VNI 4096와 UDP 4789 포트 사용은 제한된다. VNI 제한은 작업 중이며 향후 릴리스(오픈 소스 flannel 변경)에서 구현될 것이다. 이러한 파라미터에 대한 자세한 내용은 공식 [Flannel VXLAN](https://github.com/coreos/flannel/blob/master/Documentation/backends.md#vxlan) 백엔드 문서를 참고한다. + 1. 노드-파드 연결은 설계상 불가능하다. Flannel v0.12.0(또는 그 이상)이 + 있는 로컬 파드에서만 가능하다. + + 1. VNI 4096와 UDP 4789 포트 사용은 제한된다. VNI 제한은 + 작업 중이며 향후 릴리스(오픈 소스 flannel 변경)에서 + 구현될 것이다. 이러한 파라미터에 대한 자세한 내용은 공식 + [Flannel VXLAN](https://github.com/coreos/flannel/blob/master/Documentation/backends.md#vxlan) + 백엔드 문서를 참고한다. ##### DNS {#dns-limitations} -* ClusterFirstWithHostNet은 DNS에서 지원되지 않는다. 윈도우는 '.'이 있는 모든 이름을 FQDN으로 처리하고 PQDN 확인을 건너뛴다. -* 리눅스에서는 PQDN을 확인하려고 할 때 사용되는 DNS 접미사 목록이 있다. 윈도우에서는 해당 파드의 네임스페이스(예: mydns.svc.cluster.local)와 연결된 DNS 접미사인 DNS 접미사 1개만 있다. 윈도우는 FQDN과 서비스 또는 해당 접미사만으로 확인할 수 있는 이름을 확인할 수 있다. 예를 들어, 디폴트 네임스페이스에서 생성된 파드에는 DNS 접미사 **default.svc.cluster.local**이 있다. 윈도우 파드에서는 **kubernetes.default.svc.cluster.local** 및 **kubernetes**를 모두 확인할 수 있지만 **kubernetes.default** 또는 **kubernetes.default.svc**와 같은 중간 항목은 확인할 수 없다. -* 윈도우에서는 사용할 수 있는 여러 가지의 DNS 리졸버(resolver)가 있다. 이들은 약간 다른 동작을 제공하므로, 이름 쿼리 확인을 위해 `Resolve-DNSName` 유틸리티를 사용하는 것이 좋다. +* ClusterFirstWithHostNet은 DNS에서 지원되지 않는다. 윈도우는 + '.'이 있는 모든 이름을 FQDN으로 처리하고 PQDN 확인을 건너뛴다. + +* 리눅스에서는 PQDN을 확인하려고 할 때 사용되는 DNS 접미사 목록이 + 있다. 윈도우에서는 해당 파드의 네임스페이스(예: mydns.svc.cluster.local)와 + 연결된 DNS 접미사인 DNS 접미사 1개만 있다. + 윈도우는 FQDN과 서비스 또는 해당 접미사만으로 확인할 수 있는 이름을 확인할 수 + 있다. 예를 들어, 디폴트 네임스페이스에서 생성된 파드에는 DNS + 접미사 `default.svc.cluster.local`이 있다. 윈도우 파드에서는 + `kubernetes.default.svc.cluster.local` 및 `kubernetes`를 모두 확인할 수 + 있지만 `kubernetes.default` 또는 `kubernetes.default.svc`와 같은 중간 항목은 확인할 수 없다. + +* 윈도우에서는 사용할 수 있는 여러 가지의 DNS 리졸버(resolver)가 있다. 이들은 + 약간 다른 동작을 제공하므로, 이름 쿼리 확인을 위해 `Resolve-DNSName` 유틸리티를 + 사용하는 것이 좋다. ##### IPv6 -윈도우의 쿠버네티스는 단일 스택 "IPv6 전용" 네트워킹을 지원하지 않는다. 그러나 단일 제품군 서비스를 사용하는 파드와 노드에 대한 이중 스택 IPv4/IPv6 네트워킹이 지원된다. 자세한 내용은 [IPv4/IPv6 이중 스택 네트워킹](#ipv4ipv6-이중-스택)을 참고한다. +윈도우의 쿠버네티스는 단일 스택 "IPv6 전용" 네트워킹을 지원하지 않는다. +그러나 단일 제품군 서비스를 사용하는 파드와 노드에 대한 이중 스택 IPv4/IPv6 네트워킹이 +지원된다. +자세한 내용은 [IPv4/IPv6 이중 스택 네트워킹](#ipv4ipv6-이중-스택)을 참고한다. ##### 세션 어피니티(affinity) -`service.spec.sessionAffinityConfig.clientIP.timeoutSeconds`를 사용하는 윈도우 서비스의 최대 세션 고정(sticky) 시간 설정은 지원되지 않는다. +`service.spec.sessionAffinityConfig.clientIP.timeoutSeconds`를 사용하는 +윈도우 서비스의 최대 세션 고정(sticky) 시간 설정은 지원되지 않는다. ##### 보안 -시크릿(Secret)은 노드의 볼륨(리눅스의 tmpfs/in-memory와 비교)에 일반 텍스트로 작성된다. 이는 고객이 두 가지 작업을 수행해야 함을 의미한다. +시크릿(Secret)은 노드의 볼륨(리눅스의 tmpfs/in-memory와 +비교)에 일반 텍스트로 작성된다. 이는 고객이 두 가지 작업을 수행해야 함을 의미한다. 1. 파일 ACL을 사용하여 시크릿 파일 위치를 보호한다. -2. [BitLocker](https://docs.microsoft.com/ko-kr/windows/security/information-protection/bitlocker/bitlocker-how-to-deploy-on-windows-server)를 사용한 볼륨-레벨 암호화를 사용한다. +1. [BitLocker](https://docs.microsoft.com/ko-kr/windows/security/information-protection/bitlocker/bitlocker-how-to-deploy-on-windows-server)를 + 사용한 볼륨-레벨 암호화를 사용한다. -[RunAsUsername](/ko/docs/tasks/configure-pod-container/configure-runasusername)은 컨테이너 프로세스를 노드 기본 사용자로 실행하기 위해 윈도우 파드 또는 컨테이너에 지정할 수 있다. 이것은 [RunAsUser](/ko/docs/concepts/policy/pod-security-policy/#사용자-및-그룹)와 거의 동일하다. +[RunAsUsername](/ko/docs/tasks/configure-pod-container/configure-runasusername)은 +컨테이너 프로세스를 노드 기본 사용자로 실행하기 위해 윈도우 파드 또는 +컨테이너에 지정할 수 있다. 이것은 +[RunAsUser](/ko/docs/concepts/policy/pod-security-policy/#사용자-및-그룹)와 거의 동일하다. -SELinux, AppArmor, Seccomp, 기능(POSIX 기능)과 같은 리눅스 특유의 파드 시큐리티 컨텍스트 권한은 지원하지 않는다. +SELinux, AppArmor, Seccomp, 기능(POSIX 기능)과 같은 +리눅스 특유의 파드 시큐리티 컨텍스트 권한은 지원하지 않는다. -또한 이미 언급했듯이 특권을 가진 컨테이너는 윈도우에서 지원되지 않는다. +또한 이미 언급했듯이 특권을 가진 컨테이너는 윈도우에서 지원되지 +않는다. #### API -대부분의 Kubernetes API가 윈도우에서 작동하는 방식은 차이가 없다. 중요한 차이점은 OS와 컨테이너 런타임의 차이로 귀결된다. 특정 상황에서 파드 또는 컨테이너와 같은 워크로드 API의 일부 속성은 리눅스에서 구현되고 윈도우에서 실행되지 않는다는 가정 하에 설계되었다. +대부분의 Kubernetes API가 윈도우에서 작동하는 방식은 차이가 없다. +중요한 차이점은 OS와 컨테이너 런타임의 차이로 +귀결된다. 특정 상황에서 파드 또는 컨테이너와 같은 워크로드 API의 +일부 속성은 리눅스에서 구현되고 윈도우에서 실행되지 않는다는 가정 하에 +설계되었다. 높은 수준에서 이러한 OS 개념은 다르다. -* ID - 리눅스는 정수형으로 표시되는 userID(UID) 및 groupID(GID)를 사용한다. 사용자와 그룹 이름은 정식 이름이 아니다. UID+GID에 대한 `/etc/groups` 또는 `/etc/passwd`의 별칭일 뿐이다. 윈도우는 윈도우 보안 계정 관리자(Security Account Manager, SAM) 데이터베이스에 저장된 더 큰 이진 보안 식별자(SID)를 사용한다. 이 데이터베이스는 호스트와 컨테이너 간에 또는 컨테이너들 간에 공유되지 않는다. -* 파일 퍼미션 - 윈도우는 권한 및 UUID+GID의 비트 마스크(bitmask) 대신 SID를 기반으로 하는 접근 제어 목록을 사용한다. -* 파일 경로 - 윈도우의 규칙은 `/` 대신 `\`를 사용하는 것이다. Go IO 라이브러리는 두 가지 파일 경로 분리자를 모두 허용한다. 하지만, 컨테이너 내부에서 해석되는 경로 또는 커맨드 라인을 설정할 때 `\`가 필요할 수 있다. -* 신호(Signals) - 윈도우 대화형(interactive) 앱은 종료를 다르게 처리하며, 다음 중 하나 이상을 구현할 수 있다. - * UI 스레드는 WM_CLOSE를 포함하여 잘 정의된(well-defined) 메시지를 처리한다. - * 콘솔 앱은 컨트롤 핸들러(Control Handler)를 사용하여 ctrl-c 또는 ctrl-break를 처리한다. - * 서비스는 SERVICE_CONTROL_STOP 제어 코드를 수용할 수 있는 Service Control Handler 함수를 등록한다. +* ID - 리눅스는 정수형으로 표시되는 userID(UID) 및 groupID(GID)를 + 사용한다. 사용자와 그룹 이름은 정식 이름이 아니다. UID+GID에 대한 + `/etc/groups` 또는 `/etc/passwd`의 별칭일 뿐이다. 윈도우는 윈도우 + 보안 계정 관리자(Security Account Manager, SAM) 데이터베이스에 + 저장된 더 큰 이진 보안 식별자(SID)를 사용한다. 이 데이터베이스는 호스트와 + 컨테이너 간에 또는 컨테이너들 간에 공유되지 않는다. -종료 코드는 0일 때 성공, 0이 아닌 경우 실패인 동일한 규칙을 따른다. 특정 오류 코드는 윈도우와 리눅스에서 다를 수 있다. 그러나 쿠버네티스 컴포넌트(kubelet, kube-proxy)에서 전달된 종료 코드는 변경되지 않는다. +* 파일 퍼미션 - 윈도우는 권한 및 UUID+GID의 비트 마스크(bitmask) 대신 + SID를 기반으로 하는 접근 제어 목록을 사용한다. + +* 파일 경로 - 윈도우의 규칙은 `/` 대신 `\`를 사용하는 것이다. Go IO + 라이브러리는 두 가지 파일 경로 분리자를 모두 허용한다. 하지만, 컨테이너 + 내부에서 해석되는 경로 또는 커맨드 라인을 설정할 때 `\`가 필요할 수 + 있다. + +* 신호(Signals) - 윈도우 대화형(interactive) 앱은 종료를 다르게 처리하며, 다음 중 + 하나 이상을 구현할 수 있다. + + * UI 스레드는 `WM_CLOSE`를 포함하여 잘 정의된(well-defined) 메시지를 처리한다. + + * 콘솔 앱은 컨트롤 핸들러(Control Handler)를 사용하여 ctrl-c 또는 ctrl-break를 처리한다. + + * 서비스는 `SERVICE_CONTROL_STOP` 제어 코드를 수용할 수 있는 + Service Control Handler 함수를 등록한다. + +종료 코드는 0일 때 성공, 0이 아닌 경우 실패인 동일한 규칙을 따른다. +특정 오류 코드는 윈도우와 리눅스에서 다를 수 있다. 그러나 +쿠버네티스 컴포넌트(kubelet, kube-proxy)에서 전달된 종료 코드는 +변경되지 않는다. ##### V1.Container -* V1.Container.ResourceRequirements.limits.cpu 및 V1.Container.ResourceRequirements.limits.memory - 윈도우는 CPU 할당에 하드 리밋(hard limit)을 사용하지 않는다. 대신 공유 시스템이 사용된다. 밀리코어를 기반으로 하는 기존 필드는 윈도우 스케줄러가 뒤따르는 상대적인 공유로 스케일된다. [참고: kuberuntime/helpers_windows.go](https://github.com/kubernetes/kubernetes/blob/master/pkg/kubelet/kuberuntime/helpers_windows.go), [참고: Microsoft 문서 내 리소스 제어](https://docs.microsoft.com/ko-kr/virtualization/windowscontainers/manage-containers/resource-controls) - * Huge page는 윈도우 컨테이너 런타임에서 구현되지 않으며, 사용할 수 없다. 컨테이너에 대해 구성할 수 없는 [사용자 권한(privilege) 어설트](https://docs.microsoft.com/en-us/windows/desktop/Memory/large-page-support)가 필요하다. -* V1.Container.ResourceRequirements.requests.cpu 및 V1.Container.ResourceRequirements.requests.memory - 노드의 사용 가능한 리소스에서 요청(requests)을 빼서, 노드에 대한 오버 프로비저닝을 방지하는데 사용할 수 있다. 그러나 오버 프로비저닝된 노드에서 리소스를 보장하는 데는 사용할 수 없다. 운영자가 오버 프로비저닝을 완전히 피하려는 경우 모범 사례로 모든 컨테이너에 적용해야 한다. -* V1.Container.SecurityContext.allowPrivilegeEscalation - 윈도우에서는 불가능하며, 어떤 기능도 연결되지 않는다. -* V1.Container.SecurityContext.Capabilities - POSIX 기능은 윈도우에서 구현되지 않는다. -* V1.Container.SecurityContext.privileged - 윈도우는 특권을 가진 컨테이너를 지원하지 않는다. +* V1.Container.ResourceRequirements.limits.cpu 및 + V1.Container.ResourceRequirements.limits.memory - 윈도우는 CPU 할당에 하드 + 리밋(hard limit)을 사용하지 않는다. 대신 공유 시스템이 사용된다. 밀리코어를 + 기반으로 하는 기존 필드는 윈도우 스케줄러가 뒤따르는 상대적인 공유로 + 스케일된다. + [참고: kuberuntime/helpers_windows.go](https://github.com/kubernetes/kubernetes/blob/master/pkg/kubelet/kuberuntime/helpers_windows.go), + [참고: Microsoft 문서 내 리소스 제어](https://docs.microsoft.com/ko-kr/virtualization/windowscontainers/manage-containers/resource-controls) + + * Huge page는 윈도우 컨테이너 런타임에서 구현되지 않으며, + 사용할 수 없다. 컨테이너에 대해 구성할 수 없는 + [사용자 권한(privilege) 어설트](https://docs.microsoft.com/en-us/windows/desktop/Memory/large-page-support)가 + 필요하다. + +* V1.Container.ResourceRequirements.requests.cpu 및 + V1.Container.ResourceRequirements.requests.memory - 노드의 사용 가능한 + 리소스에서 요청(requests)을 빼서, 노드에 대한 오버 프로비저닝을 방지하는데 사용할 수 + 있다. 그러나 오버 프로비저닝된 노드에서 리소스를 보장하는 데는 + 사용할 수 없다. 운영자가 오버 프로비저닝을 완전히 피하려는 경우 + 모범 사례로 모든 컨테이너에 적용해야 한다. + +* V1.Container.SecurityContext.allowPrivilegeEscalation - 윈도우에서는 + 불가능하며, 어떤 기능도 연결되지 않는다. + +* V1.Container.SecurityContext.Capabilities - POSIX 기능은 윈도우에서 + 구현되지 않는다. + +* V1.Container.SecurityContext.privileged - 윈도우는 특권을 가진 컨테이너를 + 지원하지 않는다. + * V1.Container.SecurityContext.procMount - 윈도우에는 /proc 파일시스템이 없다. -* V1.Container.SecurityContext.readOnlyRootFilesystem - 윈도우에서는 불가능하며, 레지스트리 및 시스템 프로세스가 컨테이너 내부에서 실행되려면 쓰기 권한이 필요하다. + +* V1.Container.SecurityContext.readOnlyRootFilesystem - 윈도우에서는 불가능하며, + 레지스트리 및 시스템 프로세스가 컨테이너 내부에서 실행되려면 쓰기 권한이 + 필요하다. + * V1.Container.SecurityContext.runAsGroup - 윈도우에서는 불가능하며, GID 지원이 없다. -* V1.Container.SecurityContext.runAsNonRoot - 윈도우에는 root 사용자가 없다. 가장 가까운 항목은 노드에 존재하지 않는 아이덴티티(identity)인 ContainerAdministrator이다. -* V1.Container.SecurityContext.runAsUser - 윈도우에서는 불가능하며, 정수값으로의 UID 지원이 없다. + +* V1.Container.SecurityContext.runAsNonRoot - 윈도우에는 root 사용자가 + 없다. 가장 가까운 항목은 노드에 존재하지 않는 아이덴티티(identity)인 + ContainerAdministrator이다. + +* V1.Container.SecurityContext.runAsUser - 윈도우에서는 불가능하며, 정수값으로의 UID + 지원이 없다. + * V1.Container.SecurityContext.seLinuxOptions - 윈도우에서는 불가능하며, SELinux가 없다. -* V1.Container.terminationMessagePath - 윈도우가 단일 파일 매핑을 지원하지 않는다는 점에서 몇 가지 제한이 있다. 기본값은 /dev/termination-log이며, 기본적으로 윈도우에 존재하지 않기 때문에 작동한다. + +* V1.Container.terminationMessagePath - 윈도우가 단일 파일 매핑을 지원하지 + 않는다는 점에서 몇 가지 제한이 있다. 기본값은 /dev/termination-log이며, 기본적으로 윈도우에 존재하지 않기 때문에 + 작동한다. ##### V1.Pod * V1.Pod.hostIPC, v1.pod.hostpid - 윈도우에서 호스트 네임스페이스 공유가 불가능하다. + * V1.Pod.hostNetwork - 호스트 네트워크를 공유하기 위한 윈도우 OS 지원이 없다. -* V1.Pod.dnsPolicy - ClusterFirstWithHostNet - 윈도우에서 호스트 네트워킹이 지원되지 않기 때문에 지원되지 않는다. + +* V1.Pod.dnsPolicy - ClusterFirstWithHostNet - 윈도우에서 호스트 네트워킹이 지원되지 않기 때문에 + 지원되지 않는다. + * V1.Pod.podSecurityContext - 아래 V1.PodSecurityContext 내용을 참고한다. -* V1.Pod.shareProcessNamespace - 이것은 베타 기능이며, 윈도우에서 구현되지 않은 리눅스 네임스페이스에 따라 다르다. 윈도우는 프로세스 네임스페이스 또는 컨테이너의 루트 파일시스템을 공유할 수 없다. 네트워크만 공유할 수 있다. -* V1.Pod.terminationGracePeriodSeconds - 이것은 윈도우의 도커에서 완전히 구현되지 않았다. [참조](https://github.com/moby/moby/issues/25982)의 내용을 참고한다. 현재 동작은 ENTRYPOINT 프로세스가 CTRL_SHUTDOWN_EVENT로 전송된 다음, 윈도우가 기본적으로 5초를 기다린 후, 마지막으로 정상적인 윈도우 종료 동작을 사용하여 모든 프로세스를 종료하는 것이다. 5초 기본값은 실제로 [컨테이너 내부](https://github.com/moby/moby/issues/25982#issuecomment-426441183) 윈도우 레지스트리에 있으므로 컨테이너를 빌드할 때 재정의 할 수 있다. -* V1.Pod.volumeDevices - 이것은 베타 기능이며, 윈도우에서 구현되지 않는다. 윈도우는 원시 블록 장치(raw block device)를 파드에 연결할 수 없다. -* V1.Pod.volumes - EmptyDir, 시크릿, 컨피그맵, HostPath - 모두 작동하며 TestGrid에 테스트가 있다. - * V1.emptyDirVolumeSource - 노드 기본 매체는 윈도우의 디스크이다. 윈도우에는 내장 RAM 디스크가 없기 때문에 메모리는 지원되지 않는다. + +* V1.Pod.shareProcessNamespace - 이것은 베타 기능이며, 윈도우에서 구현되지 않은 + 리눅스 네임스페이스에 따라 다르다. 윈도우는 프로세스 네임스페이스 또는 + 컨테이너의 루트 파일시스템을 공유할 수 없다. 네트워크만 공유할 수 + 있다. + +* V1.Pod.terminationGracePeriodSeconds - 이것은 윈도우의 도커에서 + 완전히 구현되지 않았다. + [참조](https://github.com/moby/moby/issues/25982)의 내용을 참고한다. 현재 동작은 + `ENTRYPOINT` 프로세스가 `CTRL_SHUTDOWN_EVENT`로 전송된 다음, 윈도우가 기본적으로 5초를 + 기다린 후, 마지막으로 정상적인 윈도우 종료 동작을 사용하여 모든 프로세스를 + 종료하는 것이다. 5초 기본값은 실제로 + [컨테이너 내부](https://github.com/moby/moby/issues/25982#issuecomment-426441183) + 윈도우 레지스트리에 있으므로 컨테이너를 빌드할 때 재정의 할 수 있다. + +* V1.Pod.volumeDevices - 이것은 베타 기능이며, 윈도우에서 구현되지 + 않는다. 윈도우는 원시 블록 장치(raw block device)를 파드에 연결할 수 없다. + +* V1.Pod.volumes - EmptyDir, 시크릿, 컨피그맵, HostPath - 모두 작동하며 + TestGrid에 테스트가 있다. + + * V1.emptyDirVolumeSource - 노드 기본 매체는 윈도우의 디스크이다. + 윈도우에는 내장 RAM 디스크가 없기 때문에 메모리는 지원되지 않는다. + * V1.VolumeMount.mountPropagation - 마운트 전파(propagation)는 윈도우에서 지원되지 않는다. ##### V1.PodSecurityContext -PodSecurityContext 필드는 윈도우에서 작동하지 않는다. 참조를 위해 여기에 나열한다. +PodSecurityContext 필드는 윈도우에서 작동하지 않는다. 참조를 위해 여기에 +나열한다. * V1.PodSecurityContext.SELinuxOptions - SELinux는 윈도우에서 사용할 수 없다. + * V1.PodSecurityContext.RunAsUser - 윈도우에서는 사용할 수 없는 UID를 제공한다. + * V1.PodSecurityContext.RunAsGroup - 윈도우에서는 사용할 수 없는 GID를 제공한다. -* V1.PodSecurityContext.RunAsNonRoot - 윈도우에는 root 사용자가 없다. 가장 가까운 항목은 노드에 존재하지 않는 아이덴티티인 ContainerAdministrator이다. + +* V1.PodSecurityContext.RunAsNonRoot - 윈도우에는 root 사용자가 없다. 가장 + 가까운 항목은 노드에 존재하지 않는 아이덴티티인 + ContainerAdministrator이다. + * V1.PodSecurityContext.SupplementalGroups - 윈도우에서는 사용할 수 없는 GID를 제공한다. -* V1.PodSecurityContext.Sysctls - 이것들은 리눅스 sysctl 인터페이스의 일부이다. 윈도우에는 이에 상응하는 것이 없다. + +* V1.PodSecurityContext.Sysctls - 이것들은 리눅스 sysctl 인터페이스의 + 일부이다. 윈도우에는 이에 상응하는 것이 없다. #### 운영 체제 버전 제한 -윈도우에는 호스트 OS 버전이 컨테이너 베이스 이미지 OS 버전과 일치해야 하는 엄격한 호환성 규칙이 있다. 윈도우 서버 2019의 컨테이너 운영 체제가 있는 윈도우 컨테이너만 지원된다. 윈도우 컨테이너 이미지 버전의 일부 이전 버전과의 호환성을 가능하게 하는 컨테이너의 Hyper-V 격리는 향후 릴리스로 계획되어 있다. +윈도우에는 호스트 OS 버전이 컨테이너 베이스 이미지 OS 버전과 일치해야 하는 +엄격한 호환성 규칙이 있다. 윈도우 서버 2019의 컨테이너 +운영 체제가 있는 윈도우 컨테이너만 지원된다. 윈도우 컨테이너 이미지 버전의 일부 +이전 버전과의 호환성을 가능하게 하는 컨테이너의 Hyper-V 격리는 +향후 릴리스로 계획되어 있다. ## 도움 받기 및 트러블슈팅 {#troubleshooting} -쿠버네티스 클러스터 트러블슈팅을 위한 기본 도움말은 이 [섹션](/docs/tasks/debug-application-cluster/troubleshooting/)에서 먼저 찾아야 한다. 이 섹션에는 몇 가지 추가 윈도우 관련 트러블슈팅 도움말이 포함되어 있다. 로그는 쿠버네티스에서 트러블슈팅하는데 중요한 요소이다. 다른 기여자로부터 트러블슈팅 지원을 구할 때마다 이를 포함해야 한다. SIG-Windows [로그 수집에 대한 기여 가이드](https://github.com/kubernetes/community/blob/master/sig-windows/CONTRIBUTING.md#gathering-logs)의 지침을 따른다. +쿠버네티스 클러스터 트러블슈팅을 위한 기본 +도움말은 이 +[섹션](/docs/tasks/debug-application-cluster/troubleshooting/)에서 먼저 찾아야 한다. 이 +섹션에는 몇 가지 추가 윈도우 관련 트러블슈팅 도움말이 포함되어 있다. +로그는 쿠버네티스에서 트러블슈팅하는데 중요한 요소이다. 다른 +기여자로부터 트러블슈팅 지원을 구할 때마다 이를 포함해야 +한다. SIG-Windows +[로그 수집에 대한 기여 가이드](https://github.com/kubernetes/community/blob/master/sig-windows/CONTRIBUTING.md#gathering-logs)의 지침을 따른다. -1. start.ps1이 성공적으로 완료되었는지 어떻게 알 수 있는가? +* start.ps1이 성공적으로 완료되었는지 어떻게 알 수 있는가? - kubelet, kube-proxy 및 (Flannel을 네트워킹 솔루션으로 선택한 경우) 노드에서 실행 중인 flanneld 호스트 에이전트 프로세스를 확인할 수 있어야 하는데, 별도의 PowerShell 윈도우에서 실행 중인 로그가 표시된다. 또한 윈도우 노드는 쿠버네티스 클러스터에서 "Ready"로 조회되어야 한다. + kubelet, kube-proxy 및 (Flannel을 네트워킹 솔루션으로 + 선택한 경우) 노드에서 실행 중인 flanneld 호스트 에이전트 프로세스를 + 확인할 수 있어야 하는데, 별도의 PowerShell 윈도우에서 실행 중인 로그가 표시된다. 또한 + 윈도우 노드는 쿠버네티스 클러스터에서 "Ready"로 조회되어야 + 한다. -1. 백그라운드에서 서비스로 실행되도록 쿠버네티스 노드 프로세스를 구성할 수 있는가? +* 백그라운드에서 서비스로 실행되도록 쿠버네티스 노드 프로세스를 구성할 수 있는가? - Kubelet 및 kube-proxy는 이미 기본 윈도우 서비스로 실행되도록 구성되어 있으며, 실패(예: 프로세스 충돌) 시 서비스를 자동으로 다시 시작하여 복원력(resiliency)을 제공한다. 이러한 노드 컴포넌트를 서비스로 구성하기 위한 두 가지 옵션이 있다. + Kubelet 및 kube-proxy는 이미 기본 윈도우 서비스로 실행되도록 + 구성되어 있으며, 실패(예: 프로세스 충돌) 시 서비스를 + 자동으로 다시 시작하여 복원력(resiliency)을 + 제공한다. 이러한 노드 컴포넌트를 서비스로 구성하기 위한 + 두 가지 옵션이 있다. - 1. 네이티브 윈도우 서비스 + * 네이티브 윈도우 서비스 - Kubelet와 kube-proxy는 `sc.exe`를 사용하여 네이티브 윈도우 서비스로 실행될 수 있다. - - ```powershell - # 두 개의 개별 명령으로 kubelet 및 kube-proxy에 대한 서비스 생성 - sc.exe create <컴포넌트_명> binPath= "<바이너리_경로> --service <다른_인자>" - - # 인자에 공백이 포함된 경우 이스케이프 되어야 한다. - sc.exe create kubelet binPath= "C:\kubelet.exe --service --hostname-override 'minion' <다른_인자>" - - # 서비스 시작 - Start-Service kubelet - Start-Service kube-proxy - - # 서비스 중지 - Stop-Service kubelet (-Force) - Stop-Service kube-proxy (-Force) - - # 서비스 상태 질의 - Get-Service kubelet - Get-Service kube-proxy - ``` - - 1. nssm.exe 사용 - - 또한 언제든지 [nssm.exe](https://nssm.cc/)와 같은 대체 서비스 관리자를 사용하여 백그라운드에서 이러한 프로세스(flanneld, kubelet, kube-proxy)를 실행할 수 있다. 이 [샘플 스크립트](https://github.com/Microsoft/SDN/tree/master/Kubernetes/flannel/register-svc.ps1)를 사용하여 백그라운드에서 윈도우 서비스로 실행하기 위해 nssm.exe를 활용하여 kubelet, kube-proxy, flanneld.exe를 등록할 수 있다. - - ```powershell - register-svc.ps1 -NetworkMode <네트워크 모드> -ManagementIP <윈도우 노드 IP> -ClusterCIDR <클러스터 서브넷> -KubeDnsServiceIP -LogDir <로그 위치 디렉터리> - - # NetworkMode = 네트워크 모드 l2bridge(flannel host-gw, 기본값이기도 함) 또는 네트워크 솔루션으로 선택한 오버레이(flannel vxlan) - # ManagementIP = 윈도우 노드에 할당된 IP 주소. ipconfig를 사용하여 찾을 수 있다. - # ClusterCIDR = 클러스터 서브넷 범위. (기본값 10.244.0.0/16) - # KubeDnsServiceIP = 쿠버네티스 DNS 서비스 IP (기본값 10.96.0.10) - # LogDir = kubelet 및 kube-proxy 로그가 각각의 출력 파일로 리다이렉션되는 디렉터리(기본값 C:\k) - ``` - - 위에 언급된 스크립트가 적합하지 않은 경우, 다음 예제를 사용하여 nssm.exe를 수동으로 구성할 수 있다. - ```powershell - # flanneld.exe 등록 - nssm install flanneld C:\flannel\flanneld.exe - nssm set flanneld AppParameters --kubeconfig-file=c:\k\config --iface= --ip-masq=1 --kube-subnet-mgr=1 - nssm set flanneld AppEnvironmentExtra NODE_NAME= - nssm set flanneld AppDirectory C:\flannel - nssm start flanneld - - # kubelet.exe 등록 - # Microsoft는 mcr.microsoft.com/oss/kubernetes/pause:1.4.1에서 pause 인프라 컨테이너를 릴리스했다. - nssm install kubelet C:\k\kubelet.exe - nssm set kubelet AppParameters --hostname-override= --v=6 --pod-infra-container-image=mcr.microsoft.com/oss/kubernetes/pause:1.4.1 --resolv-conf="" --allow-privileged=true --enable-debugging-handlers --cluster-dns= --cluster-domain=cluster.local --kubeconfig=c:\k\config --hairpin-mode=promiscuous-bridge --image-pull-progress-deadline=20m --cgroups-per-qos=false --log-dir= --logtostderr=false --enforce-node-allocatable="" --network-plugin=cni --cni-bin-dir=c:\k\cni --cni-conf-dir=c:\k\cni\config - nssm set kubelet AppDirectory C:\k - nssm start kubelet - - # kube-proxy.exe 등록 (l2bridge / host-gw) - nssm install kube-proxy C:\k\kube-proxy.exe - nssm set kube-proxy AppDirectory c:\k - nssm set kube-proxy AppParameters --v=4 --proxy-mode=kernelspace --hostname-override=--kubeconfig=c:\k\config --enable-dsr=false --log-dir= --logtostderr=false - nssm.exe set kube-proxy AppEnvironmentExtra KUBE_NETWORK=cbr0 - nssm set kube-proxy DependOnService kubelet - nssm start kube-proxy - - # kube-proxy.exe 등록 (overlay / vxlan) - nssm install kube-proxy C:\k\kube-proxy.exe - nssm set kube-proxy AppDirectory c:\k - nssm set kube-proxy AppParameters --v=4 --proxy-mode=kernelspace --feature-gates="WinOverlay=true" --hostname-override= --kubeconfig=c:\k\config --network-name=vxlan0 --source-vip= --enable-dsr=false --log-dir= --logtostderr=false - nssm set kube-proxy DependOnService kubelet - nssm start kube-proxy - ``` - - - 초기 트러블슈팅을 위해 [nssm.exe](https://nssm.cc/)에서 다음 플래그를 사용하여 stdout 및 stderr을 출력 파일로 리다이렉션할 수 있다. - - ```powershell - nssm set AppStdout C:\k\mysvc.log - nssm set AppStderr C:\k\mysvc.log - ``` - - 자세한 내용은 공식 [nssm 사용](https://nssm.cc/usage) 문서를 참고한다. - -1. 내 윈도우 파드에 네트워크 연결이 없다. - - 가상 머신을 사용하는 경우, 모든 VM 네트워크 어댑터에서 MAC 스푸핑이 활성화되어 있는지 확인한다. - -1. 내 윈도우 파드가 외부 리소스를 ping 할 수 없다. - - 윈도우 파드에는 현재 ICMP 프로토콜용으로 프로그래밍된 아웃바운드 규칙이 없다. 그러나 TCP/UDP는 지원된다. 클러스터 외부 리소스에 대한 연결을 시연하려는 경우, `ping `를 해당 `curl `명령으로 대체한다. - - 여전히 문제가 발생하는 경우, [cni.conf](https://github.com/Microsoft/SDN/blob/master/Kubernetes/flannel/l2bridge/cni/config/cni.conf)의 네트워크 구성에 특별히 추가 확인이 필요하다. 언제든지 이 정적 파일을 편집할 수 있다. 구성 업데이트는 새로 생성된 모든 쿠버네티스 리소스에 적용된다. - - 쿠버네티스 네트워킹 요구 사항 중 하나([쿠버네티스 모델](/ko/docs/concepts/cluster-administration/networking/))는 클러스터 통신이 내부적으로 NAT 없이 발생하는 것이다. 이 요구 사항을 준수하기 위해 아웃바운드 NAT가 발생하지 않도록 하는 모든 통신에 대한 [ExceptionList](https://github.com/Microsoft/SDN/blob/master/Kubernetes/flannel/l2bridge/cni/config/cni.conf#L20)가 있다. 그러나 이것은 쿼리하려는 외부 IP를 ExceptionList에서 제외해야 함도 의미한다. 그래야만 윈도우 파드에서 발생하는 트래픽이 제대로 SNAT 되어 외부에서 응답을 받는다. 이와 관련하여 `cni.conf`의 ExceptionList는 다음과 같아야 한다. - - ```conf - "ExceptionList": [ - "10.244.0.0/16", # 클러스터 서브넷 - "10.96.0.0/12", # 서비스 서브넷 - "10.127.130.0/24" # 관리(호스트) 서브넷 - ] - ``` - -1. 내 윈도우 노드가 NodePort 서비스에 접근할 수 없다. - - 노드 자체에서는 로컬 NodePort 접근이 실패한다. 이것은 알려진 제약사항이다. NodePort 접근은 다른 노드 또는 외부 클라이언트에서는 가능하다. - -1. 컨테이너의 vNIC 및 HNS 엔드포인트가 삭제되었다. - - 이 문제는 `hostname-override` 파라미터가 [kube-proxy](/ko/docs/reference/command-line-tools-reference/kube-proxy/)에 전달되지 않은 경우 발생할 수 있다. 이를 해결하려면 사용자는 다음과 같이 hostname을 kube-proxy에 전달해야 한다. + Kubelet와 kube-proxy는 `sc.exe`를 사용하여 네이티브 윈도우 서비스로 실행될 수 있다. ```powershell - C:\k\kube-proxy.exe --hostname-override=$(hostname) + # 두 개의 개별 명령으로 kubelet 및 kube-proxy에 대한 서비스 생성 + sc.exe create <컴포넌트_명> binPath= "<바이너리_경로> --service <다른_인자>" + + # 인자에 공백이 포함된 경우 이스케이프 되어야 한다. + sc.exe create kubelet binPath= "C:\kubelet.exe --service --hostname-override 'minion' <다른_인자>" + + # 서비스 시작 + Start-Service kubelet + Start-Service kube-proxy + + # 서비스 중지 + Stop-Service kubelet (-Force) + Stop-Service kube-proxy (-Force) + + # 서비스 상태 질의 + Get-Service kubelet + Get-Service kube-proxy ``` -1. 플란넬(flannel)을 사용하면 클러스터에 다시 조인(join)한 후 노드에 이슈가 발생한다. + * nssm.exe 사용 - 이전에 삭제된 노드가 클러스터에 다시 조인될 때마다, flannelD는 새 파드 서브넷을 노드에 할당하려고 한다. 사용자는 다음 경로에서 이전 파드 서브넷 구성 파일을 제거해야 한다. + 또한 언제든지 [nssm.exe](https://nssm.cc/)와 같은 + 대체 서비스 관리자를 사용하여 백그라운드에서 이러한 프로세스(flanneld, kubelet, + kube-proxy)를 실행할 수 있다. 이 + [샘플 스크립트](https://github.com/Microsoft/SDN/tree/master/Kubernetes/flannel/register-svc.ps1)를 사용하여 + 백그라운드에서 윈도우 서비스로 실행하기 위해 `nssm.exe`를 활용하여 kubelet, kube-proxy, + `flanneld.exe`를 등록할 수 있다. ```powershell - Remove-Item C:\k\SourceVip.json - Remove-Item C:\k\SourceVipRequest.json + register-svc.ps1 -NetworkMode <네트워크 모드> -ManagementIP <윈도우 노드 IP> -ClusterCIDR <클러스터 서브넷> -KubeDnsServiceIP -LogDir <로그 위치 디렉터리> ``` + 파라미터 설명은 아래와 같다. -1. `start.ps1`을 시작한 후, flanneld가 "Waiting for the Network to be created"에서 멈춘다. + - `NetworkMode`: 네트워크 모드 l2bridge(flannel host-gw, + 기본값이기도 함) 또는 네트워크 솔루션으로 선택한 오버레이(flannel vxlan) + - `ManagementIP`: 윈도우 노드에 할당된 IP 주소. `ipconfig`를 사용하여 + 찾을 수 있다. + - `ClusterCIDR`: 클러스터 서브넷 범위. (기본값 10.244.0.0/16) + - `KubeDnsServiceIP`: 쿠버네티스 DNS 서비스 IP (기본값 10.96.0.10) + - `LogDir`: kubelet 및 kube-proxy 로그가 각각의 출력 파일로 + 리다이렉션되는 디렉터리(기본값 C:\k) - 이 [이슈](https://github.com/coreos/flannel/issues/1066)에 대한 수많은 보고가 있다. 플란넬 네트워크의 관리 IP가 설정될 때의 타이밍 이슈일 가능성이 높다. 해결 방법은 start.ps1을 다시 시작하거나 다음과 같이 수동으로 다시 시작하는 것이다. + 위에 언급된 스크립트가 적합하지 않은 경우, 다음 예제를 사용하여 + `nssm.exe`를 수동으로 구성할 수 있다. + + flanneld.exe를 등록한다. ```powershell - PS C:> [Environment]::SetEnvironmentVariable("NODE_NAME", "") - PS C:> C:\flannel\flanneld.exe --kubeconfig-file=c:\k\config --iface= --ip-masq=1 --kube-subnet-mgr=1 + nssm install flanneld C:\flannel\flanneld.exe + nssm set flanneld AppParameters --kubeconfig-file=c:\k\config --iface= --ip-masq=1 --kube-subnet-mgr=1 + nssm set flanneld AppEnvironmentExtra NODE_NAME= + nssm set flanneld AppDirectory C:\flannel + nssm start flanneld ``` -1. `/run/flannel/subnet.env` 누락으로 인해 윈도우 파드를 시작할 수 없다. - - 이것은 플란넬이 제대로 실행되지 않았음을 나타낸다. flanneld.exe를 다시 시작하거나 쿠버네티스 마스터의 `/run/flannel/subnet.env`에서 윈도우 워커 노드의 `C:\run\flannel\subnet.env`로 파일을 수동으로 복사할 수 있고, `FLANNEL_SUBNET` 행을 다른 숫자로 수정한다. 예를 들어, 다음은 노드 서브넷 10.244.4.1/24가 필요한 경우이다. - - ```env - FLANNEL_NETWORK=10.244.0.0/16 - FLANNEL_SUBNET=10.244.4.1/24 - FLANNEL_MTU=1500 - FLANNEL_IPMASQ=true - ``` - -1. 내 윈도우 노드가 서비스 IP를 사용하여 내 서비스에 접근할 수 없다. - - 이는 윈도우에서 현재 네트워킹 스택의 알려진 제약 사항이다. 그러나 윈도우 파드는 서비스 IP에 접근할 수 있다. - -1. kubelet을 시작할 때 네트워크 어댑터를 찾을 수 없다. - - 윈도우 네트워킹 스택에는 쿠버네티스 네트워킹이 작동하기 위한 가상 어댑터가 필요하다. 다음 명령이 (어드민 셸에서) 결과를 반환하지 않으면, Kubelet이 작동하는데 필요한 필수 구성 요소인 가상 네트워크 생성이 실패한 것이다. + kubelet.exe를 등록한다. ```powershell - Get-HnsNetwork | ? Name -ieq "cbr0" - Get-NetAdapter | ? Name -Like "vEthernet (Ethernet*" + # Microsoft는 mcr.microsoft.com/oss/kubernetes/pause:1.4.1에서 pause 인프라 컨테이너를 릴리스했다. + nssm install kubelet C:\k\kubelet.exe + nssm set kubelet AppParameters --hostname-override= --v=6 --pod-infra-container-image=mcr.microsoft.com/oss/kubernetes/pause:1.4.1 --resolv-conf="" --allow-privileged=true --enable-debugging-handlers --cluster-dns= --cluster-domain=cluster.local --kubeconfig=c:\k\config --hairpin-mode=promiscuous-bridge --image-pull-progress-deadline=20m --cgroups-per-qos=false --log-dir= --logtostderr=false --enforce-node-allocatable="" --network-plugin=cni --cni-bin-dir=c:\k\cni --cni-conf-dir=c:\k\cni\config + nssm set kubelet AppDirectory C:\k + nssm start kubelet ``` - 호스트 네트워크 어댑터가 "Ethernet"이 아닌 경우, 종종 start.ps1 스크립트의 [InterfaceName](https://github.com/microsoft/SDN/blob/master/Kubernetes/flannel/start.ps1#L7) 파라미터를 수정하는 것이 좋다. 그렇지 않으면 `start-kubelet.ps1` 스크립트의 출력을 참조하여 가상 네트워크 생성 중에 오류가 있는지 확인한다. + kube-proxy.exe를 등록한다(l2bridge / host-gw). -1. 내 파드가 "Container Creating"에서 멈췄거나 계속해서 다시 시작된다. - - pause 이미지가 OS 버전과 호환되는지 확인한다. [지침](https://docs.microsoft.com/en-us/virtualization/windowscontainers/kubernetes/deploying-resources)에서는 OS와 컨테이너가 모두 버전 1803이라고 가정한다. 이후 버전의 윈도우가 있는 경우, Insider 빌드와 같이 그에 따라 이미지를 조정해야 한다. 이미지는 Microsoft의 [도커 리포지터리](https://hub.docker.com/u/microsoft/)를 참조한다. 그럼에도 불구하고, pause 이미지 Dockerfile과 샘플 서비스는 이미지가 :latest로 태그될 것으로 예상한다. - -1. DNS 확인(resolution)이 제대로 작동하지 않는다. - - 이 [섹션](#dns-limitations)에서 윈도우에 대한 DNS 제한을 확인한다. - -1. `kubectl port-forward`가 "unable to do port forwarding: wincat not found"로 실패한다. - - 이는 쿠버네티스 1.15 및 pause 인프라 컨테이너 `mcr.microsoft.com/oss/kubernetes/pause:1.4.1`에서 구현되었다. 해당 버전 또는 최신 버전을 사용해야 한다. - 자체 pause 인프라 컨테이너를 빌드하려면 [wincat](https://github.com/kubernetes-sigs/sig-windows-tools/tree/master/cmd/wincat)을 포함해야 한다. - -1. 내 윈도우 서버 노드가 프록시 뒤에 있기 때문에 내 쿠버네티스 설치가 실패한다. - - 프록시 뒤에 있는 경우 다음 PowerShell 환경 변수를 정의해야 한다. - - ```PowerShell - [Environment]::SetEnvironmentVariable("HTTP_PROXY", "http://proxy.example.com:80/", [EnvironmentVariableTarget]::Machine) - [Environment]::SetEnvironmentVariable("HTTPS_PROXY", "http://proxy.example.com:443/", [EnvironmentVariableTarget]::Machine) + ```powershell + nssm install kube-proxy C:\k\kube-proxy.exe + nssm set kube-proxy AppDirectory c:\k + nssm set kube-proxy AppParameters --v=4 --proxy-mode=kernelspace --hostname-override=--kubeconfig=c:\k\config --enable-dsr=false --log-dir= --logtostderr=false + nssm.exe set kube-proxy AppEnvironmentExtra KUBE_NETWORK=cbr0 + nssm set kube-proxy DependOnService kubelet + nssm start kube-proxy ``` -1. `pause` 컨테이너란 무엇인가? + kube-proxy.exe를 등록한다(overlay / vxlan). - 쿠버네티스 파드에서는 컨테이너 엔드포인트를 호스팅하기 위해 먼저 인프라 또는 "pause" 컨테이너가 생성된다. 인프라 및 워커 컨테이너를 포함하여 동일한 파드에 속하는 컨테이너는 공통 네트워크 네임스페이스 및 엔드포인트(동일한 IP 및 포트 공간)를 공유한다. 네트워크 구성을 잃지 않고 워커 컨테이너가 충돌하거나 다시 시작되도록 하려면 pause 컨테이너가 필요하다. + ```powershell + nssm install kube-proxy C:\k\kube-proxy.exe + nssm set kube-proxy AppDirectory c:\k + nssm set kube-proxy AppParameters --v=4 --proxy-mode=kernelspace --feature-gates="WinOverlay=true" --hostname-override= --kubeconfig=c:\k\config --network-name=vxlan0 --source-vip= --enable-dsr=false --log-dir= --logtostderr=false + nssm set kube-proxy DependOnService kubelet + nssm start kube-proxy + ``` - "pause" (인프라) 이미지는 Microsoft Container Registry(MCR)에서 호스팅된다. `mcr.microsoft.com/oss/kubernetes/pause:1.4.1`을 사용하여 접근할 수 있다. 자세한 내용은 [DOCKERFILE](https://github.com/kubernetes-sigs/windows-testing/blob/master/images/pause/Dockerfile)을 참고한다. + + 초기 트러블슈팅을 위해 [nssm.exe](https://nssm.cc/)에서 + 다음 플래그를 사용하여 stdout 및 stderr을 출력 파일로 리다이렉션할 수 있다. + + ```powershell + nssm set AppStdout C:\k\mysvc.log + nssm set AppStderr C:\k\mysvc.log + ``` + + 자세한 내용은 공식 [nssm 사용](https://nssm.cc/usage) 문서를 참고한다. + +* 내 윈도우 파드에 네트워크 연결이 없다. + + 가상 머신을 사용하는 경우, 모든 VM 네트워크 어댑터에서 MAC 스푸핑이 + 활성화되어 있는지 확인한다. + +* 내 윈도우 파드가 외부 리소스를 ping 할 수 없다. + + 윈도우 파드에는 현재 ICMP 프로토콜용으로 프로그래밍된 아웃바운드 + 규칙이 없다. 그러나 TCP/UDP는 지원된다. 클러스터 외부 리소스에 대한 연결을 + 시연하려는 경우, `ping `를 해당 `curl `명령으로 + 대체한다. + + 여전히 문제가 발생하는 경우, + [cni.conf](https://github.com/Microsoft/SDN/blob/master/Kubernetes/flannel/l2bridge/cni/config/cni.conf)의 + 네트워크 구성에 특별히 추가 확인이 필요하다. 언제든지 이 정적 파일을 편집할 수 있다. 구성 + 업데이트는 새로 생성된 모든 쿠버네티스 리소스에 적용된다. + + 쿠버네티스 네트워킹 요구 사항 중 + 하나([쿠버네티스 모델](/ko/docs/concepts/cluster-administration/networking/))는 + 클러스터 통신이 내부적으로 NAT 없이 발생하는 것이다. 이 요구 사항을 + 준수하기 위해 아웃바운드 NAT가 발생하지 않도록 하는 모든 통신에 대한 + [ExceptionList](https://github.com/Microsoft/SDN/blob/master/Kubernetes/flannel/l2bridge/cni/config/cni.conf#L20)가 + 있다. 그러나 + 이것은 쿼리하려는 외부 IP를 ExceptionList에서 + 제외해야 함도 의미한다. 그래야만 윈도우 파드에서 발생하는 트래픽이 제대로 SNAT 되어 + 외부에서 응답을 받는다. 이와 관련하여 `cni.conf`의 ExceptionList는 다음과 + 같아야 한다. + + ```conf + "ExceptionList": [ + "10.244.0.0/16", # 클러스터 서브넷 + "10.96.0.0/12", # 서비스 서브넷 + "10.127.130.0/24" # 관리(호스트) 서브넷 + ] + ``` + +* 내 윈도우 노드가 NodePort 서비스에 접근할 수 없다. + + 노드 자체에서는 로컬 NodePort 접근이 실패한다. 이것은 알려진 + 제약사항이다. NodePort 접근은 다른 노드 또는 외부 클라이언트에서는 가능하다. + +* 컨테이너의 vNIC 및 HNS 엔드포인트가 삭제되었다. + + 이 문제는 `hostname-override` 파라미터가 + [kube-proxy](/ko/docs/reference/command-line-tools-reference/kube-proxy/)에 + 전달되지 않은 경우 발생할 수 있다. + 이를 해결하려면 사용자는 다음과 같이 hostname을 kube-proxy에 전달해야 한다. + + ```powershell + C:\k\kube-proxy.exe --hostname-override=$(hostname) + ``` + +* 플란넬(flannel)을 사용하면 클러스터에 다시 조인(join)한 후 노드에 이슈가 발생한다. + + 이전에 삭제된 노드가 클러스터에 다시 조인될 때마다, + flannelD는 새 파드 서브넷을 노드에 할당하려고 한다. 사용자는 다음 경로에서 + 이전 파드 서브넷 구성 파일을 제거해야 한다. + + ```powershell + Remove-Item C:\k\SourceVip.json + Remove-Item C:\k\SourceVipRequest.json + ``` + +* `start.ps1`을 시작한 후, flanneld가 "Waiting for the Network + to be created"에서 멈춘다. + + 이 [이슈](https://github.com/coreos/flannel/issues/1066)에 + 대한 수많은 보고가 있다. 플란넬 네트워크의 + 관리 IP가 설정될 때의 타이밍 이슈일 가능성이 높다. 해결 + 방법은 start.ps1을 다시 시작하거나 다음과 같이 수동으로 다시 시작하는 것이다. + + ```powershell + PS C:> [Environment]::SetEnvironmentVariable("NODE_NAME", "") + PS C:> C:\flannel\flanneld.exe --kubeconfig-file=c:\k\config --iface= --ip-masq=1 --kube-subnet-mgr=1 + ``` + +* `/run/flannel/subnet.env` 누락으로 인해 윈도우 파드를 시작할 수 없다. + + 이것은 플란넬이 제대로 실행되지 않았음을 나타낸다. flanneld.exe를 + 다시 시작하거나 쿠버네티스 마스터의 + `/run/flannel/subnet.env`에서 윈도우 워커 노드의 + `C:\run\flannel\subnet.env`로 파일을 수동으로 복사할 수 있고, + `FLANNEL_SUBNET` 행을 다른 숫자로 수정한다. 예를 들어, 다음은 노드 서브넷 + 10.244.4.1/24가 필요한 경우이다. + + ```none + FLANNEL_NETWORK=10.244.0.0/16 + FLANNEL_SUBNET=10.244.4.1/24 + FLANNEL_MTU=1500 + FLANNEL_IPMASQ=true + ``` + +* 내 윈도우 노드가 서비스 IP를 사용하여 내 서비스에 접근할 수 없다. + + 이는 윈도우에서 현재 네트워킹 스택의 알려진 제약 사항이다. + 그러나 윈도우 파드는 서비스 IP에 접근할 수 있다. + +* kubelet을 시작할 때 네트워크 어댑터를 찾을 수 없다. + + 윈도우 네트워킹 스택에는 쿠버네티스 네트워킹이 작동하기 위한 + 가상 어댑터가 필요하다. 다음 명령이 (어드민 셸에서) 결과를 반환하지 + 않으면, Kubelet이 작동하는데 필요한 필수 구성 요소인 가상 네트워크 생성이 + 실패한 것이다. + + ```powershell + Get-HnsNetwork | ? Name -ieq "cbr0" + Get-NetAdapter | ? Name -Like "vEthernet (Ethernet*" + ``` + + 호스트 네트워크 어댑터가 "Ethernet"이 아닌 경우, + 종종 start.ps1 스크립트의 + [InterfaceName](https://github.com/microsoft/SDN/blob/master/Kubernetes/flannel/start.ps1#L7) + 파라미터를 수정하는 것이 좋다. 그렇지 않으면 `start-kubelet.ps1` + 스크립트의 출력을 참조하여 가상 네트워크 생성 중에 오류가 있는지 확인한다. + +* 내 파드가 "Container Creating"에서 멈췄거나 계속해서 다시 시작된다. + + pause 이미지가 OS 버전과 호환되는지 확인한다. + [지침](https://docs.microsoft.com/en-us/virtualization/windowscontainers/kubernetes/deploying-resources)에서는 + OS와 컨테이너가 모두 버전 1803이라고 가정한다. 이후 버전의 + 윈도우가 있는 경우, Insider 빌드와 같이 그에 따라 이미지를 + 조정해야 한다. 이미지는 Microsoft의 + [도커 리포지터리](https://hub.docker.com/u/microsoft/)를 참조한다. + 그럼에도 불구하고, pause 이미지 Dockerfile과 샘플 서비스는 이미지가 + :latest로 태그될 것으로 예상한다. + +* DNS 확인(resolution)이 제대로 작동하지 않는다. + + [윈도우에 대한 DNS 제한](#dns-limitations)을 확인한다. + +* `kubectl port-forward`가 "unable to do port forwarding: wincat not found"로 실패한다. + + 이는 쿠버네티스 1.15 및 pause 인프라 컨테이너 + `mcr.microsoft.com/oss/kubernetes/pause:1.4.1`에서 구현되었다. + 해당 버전 또는 최신 버전을 사용해야 한다. 자체 pause + 인프라 컨테이너를 빌드하려면 + [wincat](https://github.com/kubernetes-sigs/sig-windows-tools/tree/master/cmd/wincat)을 포함해야 한다. + +* 내 윈도우 서버 노드가 프록시 뒤에 있기 때문에 내 쿠버네티스 + 설치가 실패한다. + + 프록시 뒤에 있는 경우 다음 PowerShell 환경 변수를 + 정의해야 한다. + + ```PowerShell + [Environment]::SetEnvironmentVariable("HTTP_PROXY", "http://proxy.example.com:80/", [EnvironmentVariableTarget]::Machine) + [Environment]::SetEnvironmentVariable("HTTPS_PROXY", "http://proxy.example.com:443/", [EnvironmentVariableTarget]::Machine) + ``` + +* `pause` 컨테이너란 무엇인가? + + 쿠버네티스 파드에서는 컨테이너 엔드포인트를 호스팅하기 위해 + 먼저 인프라 또는 "pause" 컨테이너가 생성된다. 인프라 및 워커 컨테이너를 포함하여 + 동일한 파드에 속하는 컨테이너는 공통 네트워크 네임스페이스 및 + 엔드포인트(동일한 IP 및 포트 공간)를 공유한다. 네트워크 구성을 잃지 않고 + 워커 컨테이너가 충돌하거나 다시 시작되도록 하려면 pause 컨테이너가 + 필요하다. + + "pause" (인프라) 이미지는 Microsoft Container Registry(MCR)에서 + 호스팅된다. `mcr.microsoft.com/oss/kubernetes/pause:1.4.1`을 사용하여 접근할 수 있다. + 자세한 내용은 + [DOCKERFILE](https://github.com/kubernetes-sigs/windows-testing/blob/master/images/pause/Dockerfile)을 참고한다. ### 추가 조사 -이러한 단계로 문제가 해결되지 않으면, 다음을 통해 쿠버네티스의 윈도우 노드에서 윈도우 컨테이너를 실행하는데 도움을 받을 수 있다. +이러한 단계로 문제가 해결되지 않으면, 다음을 통해 쿠버네티스의 윈도우 노드에서 +윈도우 컨테이너를 실행하는데 도움을 받을 수 있다. * 스택오버플로우 [윈도우 서버 컨테이너](https://stackoverflow.com/questions/tagged/windows-server-container) 주제 + * 쿠버네티스 공식 포럼 [discuss.kubernetes.io](https://discuss.kubernetes.io/) + * 쿠버네티스 슬랙 [#SIG-Windows Channel](https://kubernetes.slack.com/messages/sig-windows) ## 이슈 리포팅 및 기능 요청 -버그처럼 보이는 부분이 있거나 기능 요청을 하고 싶다면, [GitHub 이슈 트래킹 시스템](https://github.com/kubernetes/kubernetes/issues)을 활용한다. [GitHub](https://github.com/kubernetes/kubernetes/issues/new/choose)에서 이슈를 열고 SIG-Windows에 할당할 수 있다. 먼저 이전에 보고된 이슈 목록을 검색하고 이슈에 대한 경험을 언급하고 추가 로그를 첨부해야 한다. SIG-Windows 슬랙은 티켓을 만들기 전에 초기 지원 및 트러블슈팅 아이디어를 얻을 수 있는 좋은 방법이기도 하다. +버그처럼 보이는 부분이 있거나 기능 +요청을 하고 싶다면, +[GitHub 이슈 트래킹 시스템](https://github.com/kubernetes/kubernetes/issues)을 +활용한다. +[GitHub](https://github.com/kubernetes/kubernetes/issues/new/choose)에서 이슈를 열고 +SIG-Windows에 할당할 수 있다. 먼저 이전에 보고된 이슈 목록을 검색하고 +이슈에 대한 경험을 언급하고 추가 로그를 +첨부해야 한다. SIG-Windows 슬랙은 티켓을 만들기 전에 초기 지원 및 +트러블슈팅 아이디어를 얻을 수 있는 좋은 방법이기도 하다. -버그를 제출하는 경우, 다음과 같이 문제를 재현하는 방법에 대한 자세한 정보를 포함한다. +버그를 제출하는 경우, 다음과 같이 문제를 재현하는 방법에 대한 자세한 정보를 +포함한다. * 쿠버네티스 버전: kubectl version -* 환경 세부사항: 클라우드 공급자, OS 배포판, 네트워킹 선택 및 구성, 도커 버전 +* 환경 세부사항: 클라우드 공급자, OS 배포판, 네트워킹 선택 및 + 구성, 도커 버전 * 문제를 재현하기 위한 세부 단계 * [관련 로그](https://github.com/kubernetes/community/blob/master/sig-windows/CONTRIBUTING.md#gathering-logs) -* SIG-Windows 회원의 주의를 끌 수 있도록 `/sig windows`로 이슈에 대해 어노테이션을 달아 이슈에 sig/windows 태그를 지정한다. +* SIG-Windows 회원의 주의를 끌 수 있도록 `/sig windows`로 이슈에 대해 어노테이션을 달아 + 이슈에 sig/windows 태그를 지정한다. ## {{% heading "whatsnext" %}} -로드맵에는 많은 기능이 있다. 요약된 높은 수준의 목록이 아래에 포함되어 있지만, [로드맵 프로젝트](https://github.com/orgs/kubernetes/projects/8)를 보고 [기여](https://github.com/kubernetes/community/blob/master/sig-windows/)하여 윈도우 지원을 개선하는데 도움이 주는 것이 좋다. +로드맵에는 많은 기능이 있다. 요약된 높은 수준의 +목록이 아래에 포함되어 있지만, +[로드맵 프로젝트](https://github.com/orgs/kubernetes/projects/8)를 보고 +[기여](https://github.com/kubernetes/community/blob/master/sig-windows/)하여 +윈도우 지원을 개선하는데 도움이 주는 것이 좋다. ### Hyper-V 격리(isolation) -쿠버네티스에서 윈도우 컨테이너에 대해 다음 유스케이스를 사용하려면 Hyper-V 격리가 필요하다. +쿠버네티스에서 윈도우 컨테이너에 대해 다음 유스케이스를 사용하려면 +Hyper-V 격리가 필요하다. * 추가 보안을 위해 파드 간 하이퍼바이저 기반 격리 -* 하위 호환성을 통해 컨테이너를 다시 빌드할 필요 없이 노드에서 최신 윈도우 서버 버전을 실행할 수 있다. + +* 하위 호환성을 통해 컨테이너를 다시 빌드할 필요 없이 노드에서 + 최신 윈도우 서버 버전을 실행할 수 있다. + * 파드에 대한 특정 CPU/NUMA 설정 + * 메모리 격리 및 예약 -Hyper-V 격리 지원은 이후 릴리스에 추가되며 CRI-Containerd가 필요하다. +Hyper-V 격리 지원은 이후 릴리스에 추가되며 +CRI-Containerd가 필요하다. ### kubeadm 및 클러스터 API를 사용한 배포 Kubeadm은 사용자가 쿠버네티스 클러스터를 배포하기 위한 사실상의 표준이 되고 있다. kubeadm의 윈도우 노드 지원은 현재 작업 중이지만 -[여기](/ko/docs/tasks/administer-cluster/kubeadm/adding-windows-nodes/)에서 가이드를 사용할 수 있다. +[여기](/ko/docs/tasks/administer-cluster/kubeadm/adding-windows-nodes/)에서 +가이드를 사용할 수 있다. 또한 윈도우 노드가 적절하게 프로비저닝되도록 클러스터 API에 투자하고 있다. From a9fb2d35d0da6a522656d3c3b96c5228c77e5a81 Mon Sep 17 00:00:00 2001 From: Jihoon Seo Date: Fri, 28 May 2021 22:16:18 +0900 Subject: [PATCH 02/10] [ko] Update outdated files in dev-1.21-ko.3 (p2) --- content/ko/community/_index.html | 3 +- .../concepts/cluster-administration/addons.md | 2 +- .../concepts/extend-kubernetes/operator.md | 2 ++ .../concepts/overview/what-is-kubernetes.md | 2 +- .../working-with-objects/common-labels.md | 14 ++++---- .../working-with-objects/namespaces.md | 2 +- .../concepts/scheduling-eviction/_index.md | 36 +++++++++++++++++-- .../scheduling-eviction/kube-scheduler.md | 1 - .../scheduling-eviction/pod-overhead.md | 2 +- .../pod-priority-preemption.md | 0 .../resource-bin-packing.md | 2 +- .../scheduler-perf-tuning.md | 2 +- .../concepts/storage/persistent-volumes.md | 4 +-- .../docs/concepts/storage/storage-classes.md | 34 ++++++++++++++++-- .../concepts/workloads/pods/disruptions.md | 4 ++- .../workloads/pods/init-containers.md | 1 - .../pods/pod-topology-spread-constraints.md | 2 +- .../feature-gates.md | 14 +++++--- .../docs/reference/glossary/pod-disruption.md | 19 ++++++++++ 19 files changed, 118 insertions(+), 28 deletions(-) rename content/ko/docs/concepts/{configuration => scheduling-eviction}/pod-priority-preemption.md (100%) create mode 100644 content/ko/docs/reference/glossary/pod-disruption.md diff --git a/content/ko/community/_index.html b/content/ko/community/_index.html index 40bea7b523..0fa8d8658a 100644 --- a/content/ko/community/_index.html +++ b/content/ko/community/_index.html @@ -24,7 +24,8 @@ cid: community 비디오      토론      이벤트와 모임들      -새소식 +새소식      +릴리즈

diff --git a/content/ko/docs/concepts/cluster-administration/addons.md b/content/ko/docs/concepts/cluster-administration/addons.md index 7e67dd604f..270063a737 100644 --- a/content/ko/docs/concepts/cluster-administration/addons.md +++ b/content/ko/docs/concepts/cluster-administration/addons.md @@ -23,7 +23,7 @@ content_type: concept * [CNI-Genie](https://github.com/Huawei-PaaS/CNI-Genie)를 사용하면 쿠버네티스는 Calico, Canal, Flannel, Romana 또는 Weave와 같은 CNI 플러그인을 완벽하게 연결할 수 있다. * [Contiv](https://contiv.github.io)는 다양한 유스케이스와 풍부한 폴리시 프레임워크를 위해 구성 가능한 네트워킹(BGP를 사용하는 네이티브 L3, vxlan을 사용하는 오버레이, 클래식 L2 그리고 Cisco-SDN/ACI)을 제공한다. Contiv 프로젝트는 완전히 [오픈소스](https://github.com/contiv)이다. [인스톨러](https://github.com/contiv/install)는 kubeadm을 이용하거나, 그렇지 않은 경우에 대해서도 설치 옵션을 모두 제공한다. * [Contrail](https://www.juniper.net/us/en/products-services/sdn/contrail/contrail-networking/)은 [Tungsten Fabric](https://tungsten.io)을 기반으로 하며, 오픈소스이고, 멀티 클라우드 네트워크 가상화 및 폴리시 관리 플랫폼이다. Contrail과 Tungsten Fabric은 쿠버네티스, OpenShift, OpenStack 및 Mesos와 같은 오케스트레이션 시스템과 통합되어 있으며, 가상 머신, 컨테이너/파드 및 베어 메탈 워크로드에 대한 격리 모드를 제공한다. -* [Flannel](https://github.com/coreos/flannel/blob/master/Documentation/kubernetes.md)은 쿠버네티스와 함께 사용할 수 있는 오버레이 네트워크 제공자이다. +* [Flannel](https://github.com/flannel-io/flannel#deploying-flannel-manually)은 쿠버네티스와 함께 사용할 수 있는 오버레이 네트워크 제공자이다. * [Knitter](https://github.com/ZTE/Knitter/)는 쿠버네티스 파드에서 여러 네트워크 인터페이스를 지원하는 플러그인이다. * [Multus](https://github.com/Intel-Corp/multus-cni)는 쿠버네티스에서 SRIOV, DPDK, OVS-DPDK 및 VPP 기반 워크로드 외에 모든 CNI 플러그인(예: Calico, Cilium, Contiv, Flannel)을 지원하기 위해 쿠버네티스에서 다중 네트워크 지원을 위한 멀티 플러그인이다. * [OVN-Kubernetes](https://github.com/ovn-org/ovn-kubernetes/)는 Open vSwitch(OVS) 프로젝트에서 나온 가상 네트워킹 구현인 [OVN(Open Virtual Network)](https://github.com/ovn-org/ovn/)을 기반으로 하는 쿠버네티스용 네트워킹 제공자이다. OVN-Kubernetes는 OVS 기반 로드 밸런싱과 네트워크 폴리시 구현을 포함하여 쿠버네티스용 오버레이 기반 네트워킹 구현을 제공한다. diff --git a/content/ko/docs/concepts/extend-kubernetes/operator.md b/content/ko/docs/concepts/extend-kubernetes/operator.md index 21b3183b99..a0959f83dc 100644 --- a/content/ko/docs/concepts/extend-kubernetes/operator.md +++ b/content/ko/docs/concepts/extend-kubernetes/operator.md @@ -113,11 +113,13 @@ kubectl edit SampleDB/example-database # 일부 설정을 수동으로 변경하 {{% thirdparty-content %}} +* [Charmed Operator Framework](https://juju.is/) * [kubebuilder](https://book.kubebuilder.io/) 사용하기 * [KUDO](https://kudo.dev/) (Kubernetes Universal Declarative Operator) * 웹훅(WebHook)과 함께 [Metacontroller](https://metacontroller.app/)를 사용하여 직접 구현하기 * [오퍼레이터 프레임워크](https://operatorframework.io) +* [shell-operator](https://github.com/flant/shell-operator) ## {{% heading "whatsnext" %}} diff --git a/content/ko/docs/concepts/overview/what-is-kubernetes.md b/content/ko/docs/concepts/overview/what-is-kubernetes.md index 344c266d1e..5d2ef83d76 100644 --- a/content/ko/docs/concepts/overview/what-is-kubernetes.md +++ b/content/ko/docs/concepts/overview/what-is-kubernetes.md @@ -21,7 +21,7 @@ sitemap: 쿠버네티스는 컨테이너화된 워크로드와 서비스를 관리하기 위한 이식성이 있고, 확장가능한 오픈소스 플랫폼이다. 쿠버네티스는 선언적 구성과 자동화를 모두 용이하게 해준다. 쿠버네티스는 크고, 빠르게 성장하는 생태계를 가지고 있다. 쿠버네티스 서비스, 기술 지원 및 도구는 어디서나 쉽게 이용할 수 있다. -쿠버네티스란 명칭은 키잡이(helmsman)나 파일럿을 뜻하는 그리스어에서 유래했다. 구글이 2014년에 쿠버네티스 프로젝트를 오픈소스화했다. 쿠버네티스는 프로덕션 워크로드를 대규모로 운영하는 [15년 이상의 구글 경험](/blog/2015/04/borg-predecessor-to-kubernetes/)과 커뮤니티의 최고의 아이디어와 적용 사례가 결합되어 있다. +쿠버네티스란 명칭은 키잡이(helmsman)나 파일럿을 뜻하는 그리스어에서 유래했다. K8s라는 표기는 "K"와 "s"와 그 사이에 있는 8글자를 나타내는 약식 표기이다. 구글이 2014년에 쿠버네티스 프로젝트를 오픈소스화했다. 쿠버네티스는 프로덕션 워크로드를 대규모로 운영하는 [15년 이상의 구글 경험](/blog/2015/04/borg-predecessor-to-kubernetes/)과 커뮤니티의 최고의 아이디어와 적용 사례가 결합되어 있다. ## 여정 돌아보기 diff --git a/content/ko/docs/concepts/overview/working-with-objects/common-labels.md b/content/ko/docs/concepts/overview/working-with-objects/common-labels.md index 09f70af30c..c19ec0b3a9 100644 --- a/content/ko/docs/concepts/overview/working-with-objects/common-labels.md +++ b/content/ko/docs/concepts/overview/working-with-objects/common-labels.md @@ -32,14 +32,15 @@ kubectl과 대시보드와 같은 많은 도구들로 쿠버네티스 오브젝 레이블을 최대한 활용하려면 모든 리소스 오브젝트에 적용해야 한다. -| Key | Description | Example | Type | +| 키 | 설명 | 예시 | 타입 | | ----------------------------------- | --------------------- | -------- | ---- | | `app.kubernetes.io/name` | 애플리케이션 이름 | `mysql` | 문자열 | | `app.kubernetes.io/instance` | 애플리케이션의 인스턴스를 식별하는 고유한 이름 | `mysql-abcxzy` | 문자열 | | `app.kubernetes.io/version` | 애플리케이션의 현재 버전 (예: a semantic version, revision hash 등.) | `5.7.21` | 문자열 | | `app.kubernetes.io/component` | 아키텍처 내 구성요소 | `database` | 문자열 | | `app.kubernetes.io/part-of` | 이 애플리케이션의 전체 이름 | `wordpress` | 문자열 | -| `app.kubernetes.io/managed-by` | 애플리케이션의 작동을 관리하는데 사용되는 도구 | `helm` | 문자열 | +| `app.kubernetes.io/managed-by` | 애플리케이션의 작동을 관리하는 데 사용되는 도구 | `helm` | 문자열 | +| `app.kubernetes.io/created-by` | 이 리소스를 만든 컨트롤러/사용자 | `controller-manager` | 문자열 | 위 레이블의 실제 예시는 다음 스테이트풀셋 오브젝트를 고려한다. @@ -54,6 +55,7 @@ metadata: app.kubernetes.io/component: database app.kubernetes.io/part-of: wordpress app.kubernetes.io/managed-by: helm + app.kubernetes.io/created-by: controller-manager ``` ## 애플리케이션과 애플리케이션 인스턴스 @@ -76,7 +78,7 @@ WordPress가 여러 번 설치되어 각각 서로 다른 웹사이트를 서비 `Deployment` 와 `Service` 오브젝트를 통해 배포된 단순한 스테이트리스 서비스의 경우를 보자. 다음 두 식별자는 레이블을 가장 간단한 형태로 사용하는 방법을 나타낸다. -`Deployment` 는 애플리케이션을 실행하는 파드를 감시하는데 사용한다. +`Deployment` 는 애플리케이션을 실행하는 파드를 감시하는 데 사용한다. ```yaml apiVersion: apps/v1 kind: Deployment @@ -102,9 +104,9 @@ metadata: Helm을 이용해서 데이터베이스(MySQL)을 이용하는 웹 애플리케이션(WordPress)을 설치한 것과 같이 좀 더 복잡한 애플리케이션을 고려할 수 있다. -다음 식별자는 이 애플리케이션을 배포하는데 사용하는 오브젝트의 시작을 보여준다. +다음 식별자는 이 애플리케이션을 배포하는 데 사용하는 오브젝트의 시작을 보여준다. -WordPress를 배포하는데 다음과 같이 `Deployment` 로 시작한다. +WordPress를 배포하는 데 다음과 같이 `Deployment` 로 시작한다. ```yaml apiVersion: apps/v1 @@ -152,7 +154,7 @@ metadata: ... ``` -`Service` 는 WordPress의 일부로 MySQL을 노출하는데 이용한다. +`Service` 는 WordPress의 일부로 MySQL을 노출하는 데 이용한다. ```yaml apiVersion: v1 diff --git a/content/ko/docs/concepts/overview/working-with-objects/namespaces.md b/content/ko/docs/concepts/overview/working-with-objects/namespaces.md index ef75f4f081..049b30a1f7 100644 --- a/content/ko/docs/concepts/overview/working-with-objects/namespaces.md +++ b/content/ko/docs/concepts/overview/working-with-objects/namespaces.md @@ -35,7 +35,7 @@ weight: 30 [네임스페이스 관리자 가이드 문서](/docs/tasks/administer-cluster/namespaces/)에 기술되어 있다. {{< note >}} - 쿠버네티스 시스템 네임스페이스용으로 예약되어 있으므로, `kube-` 접두사로 네임스페이스를 생성하지 않는다. + `kube-` 접두사로 시작하는 네임스페이스는 쿠버네티스 시스템용으로 예약되어 있으므로, 사용자는 이러한 네임스페이스를 생성하지 않는다. {{< /note >}} ### 네임스페이스 조회 diff --git a/content/ko/docs/concepts/scheduling-eviction/_index.md b/content/ko/docs/concepts/scheduling-eviction/_index.md index 5cd57c3a29..5ae3f5822e 100644 --- a/content/ko/docs/concepts/scheduling-eviction/_index.md +++ b/content/ko/docs/concepts/scheduling-eviction/_index.md @@ -1,7 +1,37 @@ --- -title: "스케줄링과 축출(eviction)" +title: "스케줄링, 선점(Preemption), 축출(Eviction)" weight: 90 +content_type: concept description: > - 쿠버네티스에서, 스케줄링은 kubelet이 파드를 실행할 수 있도록 파드가 노드와 일치하는지 확인하는 것을 말한다. - 축출은 리소스가 부족한 노드에서 하나 이상의 파드를 사전에 장애로 처리하는 프로세스이다. + 쿠버네티스에서, 스케줄링은 kubelet이 파드를 실행할 수 있도록 + 파드를 노드에 할당하는 것을 말한다. + 선점은 우선순위가 높은 파드가 노드에 스케줄될 수 있도록 + 우선순위가 낮은 파드를 종료시키는 과정을 말한다. + 축출은 리소스가 부족한 노드에서 하나 이상의 파드를 사전에 종료시키는 프로세스이다. +no_list: true --- + +쿠버네티스에서, 스케줄링은 {{}}이 파드를 실행할 수 있도록 +{{}}를 +{{}}에 할당하는 것을 말한다. +선점은 {{}}가 높은 파드가 노드에 스케줄될 수 있도록 +우선순위가 낮은 파드를 종료시키는 과정을 말한다. +축출은 리소스가 부족한 노드에서 하나 이상의 파드를 사전에 종료시키는 프로세스이다. + +## 스케줄링 + +* [쿠버네티스 스케줄러](/ko/docs/concepts/scheduling-eviction/kube-scheduler/) +* [노드에 파드 할당하기](/ko/docs/concepts/scheduling-eviction/assign-pod-node/) +* [파드 오버헤드](/ko/docs/concepts/scheduling-eviction/pod-overhead/) +* [테인트(Taints)와 톨러레이션(Tolerations)](/ko/docs/concepts/scheduling-eviction/taint-and-toleration/) +* [스케줄링 프레임워크](/docs/concepts/scheduling-eviction/scheduling-framework/) +* [스케줄러 성능 튜닝](/ko/docs/concepts/scheduling-eviction/scheduler-perf-tuning/) +* [확장된 리소스를 위한 리소스 빈 패킹(bin packing)](/ko/docs/concepts/scheduling-eviction/resource-bin-packing/) + +## 파드 중단(disruption) + +{{}} + +* [파드 우선순위와 선점](/docs/concepts/scheduling-eviction/pod-priority-preemption/) +* [노드-압박 축출](/docs/concepts/scheduling-eviction/node-pressure-eviction/) +* [API를 이용한 축출](/docs/concepts/scheduling-eviction/api-eviction/) diff --git a/content/ko/docs/concepts/scheduling-eviction/kube-scheduler.md b/content/ko/docs/concepts/scheduling-eviction/kube-scheduler.md index 86e67978c2..8c17269a64 100644 --- a/content/ko/docs/concepts/scheduling-eviction/kube-scheduler.md +++ b/content/ko/docs/concepts/scheduling-eviction/kube-scheduler.md @@ -80,7 +80,6 @@ _스코어링_ 단계에서 스케줄러는 목록에 남아있는 노드의 순 1. [스케줄링 정책](/ko/docs/reference/scheduling/config/#프로파일)을 사용하면 필터링을 위한 _단정(Predicates)_ 및 스코어링을 위한 _우선순위(Priorities)_ 를 구성할 수 있다. 1. [스케줄링 프로파일](/ko/docs/reference/scheduling/config/#프로파일)을 사용하면 `QueueSort`, `Filter`, `Score`, `Bind`, `Reserve`, `Permit` 등의 다른 스케줄링 단계를 구현하는 플러그인을 구성할 수 있다. 다른 프로파일을 실행하도록 kube-scheduler를 구성할 수도 있다. - ## {{% heading "whatsnext" %}} * [스케줄러 성능 튜닝](/ko/docs/concepts/scheduling-eviction/scheduler-perf-tuning/)에 대해 읽기 diff --git a/content/ko/docs/concepts/scheduling-eviction/pod-overhead.md b/content/ko/docs/concepts/scheduling-eviction/pod-overhead.md index 0f474338d9..b0da80ceae 100644 --- a/content/ko/docs/concepts/scheduling-eviction/pod-overhead.md +++ b/content/ko/docs/concepts/scheduling-eviction/pod-overhead.md @@ -1,7 +1,7 @@ --- title: 파드 오버헤드 content_type: concept -weight: 20 +weight: 30 --- diff --git a/content/ko/docs/concepts/configuration/pod-priority-preemption.md b/content/ko/docs/concepts/scheduling-eviction/pod-priority-preemption.md similarity index 100% rename from content/ko/docs/concepts/configuration/pod-priority-preemption.md rename to content/ko/docs/concepts/scheduling-eviction/pod-priority-preemption.md diff --git a/content/ko/docs/concepts/scheduling-eviction/resource-bin-packing.md b/content/ko/docs/concepts/scheduling-eviction/resource-bin-packing.md index d11b7fe2ae..34ff6f3108 100644 --- a/content/ko/docs/concepts/scheduling-eviction/resource-bin-packing.md +++ b/content/ko/docs/concepts/scheduling-eviction/resource-bin-packing.md @@ -5,7 +5,7 @@ title: 확장된 리소스를 위한 리소스 빈 패킹(bin packing) content_type: concept -weight: 30 +weight: 80 --- diff --git a/content/ko/docs/concepts/scheduling-eviction/scheduler-perf-tuning.md b/content/ko/docs/concepts/scheduling-eviction/scheduler-perf-tuning.md index 9e049cd348..6be3e204c8 100644 --- a/content/ko/docs/concepts/scheduling-eviction/scheduler-perf-tuning.md +++ b/content/ko/docs/concepts/scheduling-eviction/scheduler-perf-tuning.md @@ -3,7 +3,7 @@ title: 스케줄러 성능 튜닝 content_type: concept -weight: 80 +weight: 100 --- diff --git a/content/ko/docs/concepts/storage/persistent-volumes.md b/content/ko/docs/concepts/storage/persistent-volumes.md index 3a85139cd2..c70b7413ad 100644 --- a/content/ko/docs/concepts/storage/persistent-volumes.md +++ b/content/ko/docs/concepts/storage/persistent-volumes.md @@ -540,11 +540,11 @@ spec: ### 접근 모드 -클레임은 특정 접근 모드로 저장소를 요청할 때 볼륨과 동일한 규칙을 사용한다. +클레임은 특정 접근 모드로 저장소를 요청할 때 [볼륨과 동일한 규칙](#접근-모드)을 사용한다. ### 볼륨 모드 -클레임은 볼륨과 동일한 규칙을 사용하여 파일시스템 또는 블록 장치로 볼륨을 사용함을 나타낸다. +클레임은 [볼륨과 동일한 규칙](#볼륨-모드)을 사용하여 파일시스템 또는 블록 장치로 볼륨을 사용함을 나타낸다. ### 리소스 diff --git a/content/ko/docs/concepts/storage/storage-classes.md b/content/ko/docs/concepts/storage/storage-classes.md index 94577ca182..0bec67ef8a 100644 --- a/content/ko/docs/concepts/storage/storage-classes.md +++ b/content/ko/docs/concepts/storage/storage-classes.md @@ -149,9 +149,9 @@ CSI | 1.14 (alpha), 1.16 (beta) ### 볼륨 바인딩 모드 `volumeBindingMode` 필드는 [볼륨 바인딩과 동적 -프로비저닝](/ko/docs/concepts/storage/persistent-volumes/#프로비저닝)의 시작 시기를 제어한다. +프로비저닝](/ko/docs/concepts/storage/persistent-volumes/#프로비저닝)의 시작 시기를 제어한다. 설정되어 있지 않으면, `Immediate` 모드가 기본으로 사용된다. -기본적으로, `Immediate` 모드는 퍼시스턴트볼륨클레임이 생성되면 볼륨 +`Immediate` 모드는 퍼시스턴트볼륨클레임이 생성되면 볼륨 바인딩과 동적 프로비저닝이 즉시 발생하는 것을 나타낸다. 토폴로지 제약이 있고 클러스터의 모든 노드에서 전역적으로 접근할 수 없는 스토리지 백엔드의 경우, 파드의 스케줄링 요구 사항에 대한 지식 없이 퍼시스턴트볼륨이 @@ -183,6 +183,36 @@ CSI | 1.14 (alpha), 1.16 (beta) 사전에 생성된 PV에서도 지원되지만, 지원되는 토폴로지 키와 예시를 보려면 해당 CSI 드라이버에 대한 문서를 본다. +{{< note >}} + `waitForFirstConsumer`를 사용한다면, 노드 어피니티를 지정하기 위해서 파드 스펙에 `nodeName`을 사용하지는 않아야 한다. + 만약 `nodeName`을 사용한다면, 스케줄러가 바이패스되고 PVC가 `pending` 상태로 있을 것이다. + + 대신, 아래와 같이 호스트네임을 이용하는 노드셀렉터를 사용할 수 있다. +{{< /note >}} + +```yaml +apiVersion: v1 +kind: Pod +metadata: + name: task-pv-pod +spec: + nodeSelector: + kubernetes.io/hostname: kube-01 + volumes: + - name: task-pv-storage + persistentVolumeClaim: + claimName: task-pv-claim + containers: + - name: task-pv-container + image: nginx + ports: + - containerPort: 80 + name: "http-server" + volumeMounts: + - mountPath: "/usr/share/nginx/html" + name: task-pv-storage +``` + ### 허용된 토폴로지 클러스터 운영자가 `WaitForFirstConsumer` 볼륨 바인딩 모드를 지정하면, 대부분의 상황에서 diff --git a/content/ko/docs/concepts/workloads/pods/disruptions.md b/content/ko/docs/concepts/workloads/pods/disruptions.md index 56244fae26..02730d4306 100644 --- a/content/ko/docs/concepts/workloads/pods/disruptions.md +++ b/content/ko/docs/concepts/workloads/pods/disruptions.md @@ -79,13 +79,15 @@ weight: 60 ([다중 영역 클러스터](/docs/setup/multiple-zones)를 이용한다면)에 애플리케이션을 분산해야 한다. -자발적 중단의 빈도는 다양하다. 기본적인 쿠버네티스 클러스터에서는 자발적인 운영 중단이 전혀 없다. +자발적 중단의 빈도는 다양하다. 기본적인 쿠버네티스 클러스터에서는 자동화된 자발적 중단은 발생하지 않는다(사용자가 지시한 자발적 중단만 발생한다). 그러나 클러스터 관리자 또는 호스팅 공급자가 자발적 중단이 발생할 수 있는 일부 부가 서비스를 운영할 수 있다. 예를 들어 노드 소프트웨어의 업데이트를 출시하는 경우 자발적 중단이 발생할 수 있다. 또한 클러스터(노드) 오토스케일링의 일부 구현에서는 단편화를 제거하고 노드의 효율을 높이는 과정에서 자발적 중단을 야기할 수 있다. 클러스터 관리자 또는 호스팅 공급자는 예측 가능한 자발적 중단 수준에 대해 문서화해야 한다. +파드 스펙 안에 [프라이어리티클래스 사용하기](/ko/docs/concepts/configuration/pod-priority-preemption/)와 같은 특정 환경설정 옵션 +또한 자발적(+ 비자발적) 중단을 유발할 수 있다. ## 파드 disruption budgets diff --git a/content/ko/docs/concepts/workloads/pods/init-containers.md b/content/ko/docs/concepts/workloads/pods/init-containers.md index b7a1241fc2..c8c7055408 100644 --- a/content/ko/docs/concepts/workloads/pods/init-containers.md +++ b/content/ko/docs/concepts/workloads/pods/init-containers.md @@ -326,6 +326,5 @@ myapp-pod 1/1 Running 0 9m ## {{% heading "whatsnext" %}} - * [초기화 컨테이너를 가진 파드 생성하기](/ko/docs/tasks/configure-pod-container/configure-pod-initialization/#초기화-컨테이너를-갖는-파드-생성) * [초기화 컨테이너 디버깅](/ko/docs/tasks/debug-application-cluster/debug-init-containers/) 알아보기 diff --git a/content/ko/docs/concepts/workloads/pods/pod-topology-spread-constraints.md b/content/ko/docs/concepts/workloads/pods/pod-topology-spread-constraints.md index 9bb8cfba78..2601f5c871 100644 --- a/content/ko/docs/concepts/workloads/pods/pod-topology-spread-constraints.md +++ b/content/ko/docs/concepts/workloads/pods/pod-topology-spread-constraints.md @@ -13,7 +13,7 @@ obsolete --> 사용자는 _토폴로지 분배 제약 조건_ 을 사용해서 지역, 영역, 노드 그리고 기타 사용자-정의 토폴로지 도메인과 같이 장애-도메인으로 설정된 클러스터에 걸쳐 파드가 분산되는 방식을 제어할 수 있다. 이를 통해 고가용성뿐만 아니라, 효율적인 리소스 활용의 목적을 이루는 데 도움이 된다. {{< note >}} -v1.19 이전 버전의 쿠버네티스에서는 파드 토폴로지 분배 제약조건을 사용하려면 +v1.18 이전 버전의 쿠버네티스에서는 파드 토폴로지 분배 제약조건을 사용하려면 [API 서버](/ko/docs/concepts/overview/components/#kube-apiserver)와 [스케줄러](/docs/reference/generated/kube-scheduler/)에서 `EvenPodsSpread`[기능 게이트](/ko/docs/reference/command-line-tools-reference/feature-gates/)를 diff --git a/content/ko/docs/reference/command-line-tools-reference/feature-gates.md b/content/ko/docs/reference/command-line-tools-reference/feature-gates.md index 6040ba514b..a658d58497 100644 --- a/content/ko/docs/reference/command-line-tools-reference/feature-gates.md +++ b/content/ko/docs/reference/command-line-tools-reference/feature-gates.md @@ -132,7 +132,7 @@ kubelet과 같은 컴포넌트의 기능 게이트를 설정하려면, 기능 | `IPv6DualStack` | `true` | 베타 | 1.21 | | | `KubeletCredentialProviders` | `false` | 알파 | 1.20 | | | `LegacyNodeRoleBehavior` | `false` | 알파 | 1.16 | 1.18 | -| `LegacyNodeRoleBehavior` | `true` | 베타 | 1.19 | | +| `LegacyNodeRoleBehavior` | `true` | 베타 | 1.19 | 1.20 | | `LocalStorageCapacityIsolation` | `false` | 알파 | 1.7 | 1.9 | | `LocalStorageCapacityIsolation` | `true` | 베타 | 1.10 | | | `LocalStorageCapacityIsolationFSQuotaMonitoring` | `false` | 알파 | 1.15 | | @@ -142,7 +142,7 @@ kubelet과 같은 컴포넌트의 기능 게이트를 설정하려면, 기능 | `NamespaceDefaultLabelName` | `true` | 베타 | 1.21 | | | `NetworkPolicyEndPort` | `false` | 알파 | 1.21 | | | `NodeDisruptionExclusion` | `false` | 알파 | 1.16 | 1.18 | -| `NodeDisruptionExclusion` | `true` | 베타 | 1.19 | | +| `NodeDisruptionExclusion` | `true` | 베타 | 1.19 | 1.20 | | `NonPreemptingPriority` | `false` | 알파 | 1.15 | 1.18 | | `NonPreemptingPriority` | `true` | 베타 | 1.19 | | | `PodDeletionCost` | `false` | 알파 | 1.21 | | @@ -164,7 +164,7 @@ kubelet과 같은 컴포넌트의 기능 게이트를 설정하려면, 기능 | `ServiceLBNodePortControl` | `false` | 알파 | 1.20 | | | `ServiceLoadBalancerClass` | `false` | 알파 | 1.21 | | | `ServiceNodeExclusion` | `false` | 알파 | 1.8 | 1.18 | -| `ServiceNodeExclusion` | `true` | 베타 | 1.19 | | +| `ServiceNodeExclusion` | `true` | 베타 | 1.19 | 1.20 | | `ServiceTopology` | `false` | 알파 | 1.17 | | | `SetHostnameAsFQDN` | `false` | 알파 | 1.19 | 1.19 | | `SetHostnameAsFQDN` | `true` | 베타 | 1.20 | | @@ -173,7 +173,8 @@ kubelet과 같은 컴포넌트의 기능 게이트를 설정하려면, 기능 | `StorageVersionHash` | `false` | 알파 | 1.14 | 1.14 | | `StorageVersionHash` | `true` | 베타 | 1.15 | | | `SuspendJob` | `false` | 알파 | 1.21 | | -| `TTLAfterFinished` | `false` | 알파 | 1.12 | | +| `TTLAfterFinished` | `false` | 알파 | 1.12 | 1.20 | +| `TTLAfterFinished` | `true` | 베타 | 1.21 | | | `TopologyAwareHints` | `false` | 알파 | 1.21 | | | `TopologyManager` | `false` | 알파 | 1.16 | 1.17 | | `TopologyManager` | `true` | 베타 | 1.18 | | @@ -266,6 +267,7 @@ kubelet과 같은 컴포넌트의 기능 게이트를 설정하려면, 기능 | `EvenPodsSpread` | `true` | 베타 | 1.18 | 1.18 | | `EvenPodsSpread` | `true` | GA | 1.19 | - | | `ExecProbeTimeout` | `true` | GA | 1.20 | - | +| `ExternalPolicyForExternalIP` | `true` | GA | 1.18 | - | | `GCERegionalPersistentDisk` | `true` | 베타 | 1.10 | 1.12 | | `GCERegionalPersistentDisk` | `true` | GA | 1.13 | - | | `HugePages` | `false` | 알파 | 1.8 | 1.9 | @@ -286,11 +288,13 @@ kubelet과 같은 컴포넌트의 기능 게이트를 설정하려면, 기능 | `KubeletPodResources` | `false` | 알파 | 1.13 | 1.14 | | `KubeletPodResources` | `true` | 베타 | 1.15 | | | `KubeletPodResources` | `true` | GA | 1.20 | | +| `LegacyNodeRoleBehavior` | `false` | GA | 1.21 | - | | `MountContainers` | `false` | 알파 | 1.9 | 1.16 | | `MountContainers` | `false` | 사용중단 | 1.17 | - | | `MountPropagation` | `false` | 알파 | 1.8 | 1.9 | | `MountPropagation` | `true` | 베타 | 1.10 | 1.11 | | `MountPropagation` | `true` | GA | 1.12 | - | +| `NodeDisruptionExclusion` | `true` | GA | 1.21 | - | | `NodeLease` | `false` | 알파 | 1.12 | 1.13 | | `NodeLease` | `true` | 베타 | 1.14 | 1.16 | | `NodeLease` | `true` | GA | 1.17 | - | @@ -341,6 +345,7 @@ kubelet과 같은 컴포넌트의 기능 게이트를 설정하려면, 기능 | `ServiceLoadBalancerFinalizer` | `false` | 알파 | 1.15 | 1.15 | | `ServiceLoadBalancerFinalizer` | `true` | 베타 | 1.16 | 1.16 | | `ServiceLoadBalancerFinalizer` | `true` | GA | 1.17 | - | +| `ServiceNodeExclusion` | `true` | GA | 1.21 | - | | `StartupProbe` | `false` | 알파 | 1.16 | 1.17 | | `StartupProbe` | `true` | 베타 | 1.18 | 1.19 | | `StartupProbe` | `true` | GA | 1.20 | - | @@ -636,6 +641,7 @@ kubelet과 같은 컴포넌트의 기능 게이트를 설정하려면, 기능 권한이 있는 컨테이너 또는 특정 비-네임스페이스(non-namespaced) 기능(예: `MKNODE`, `SYS_MODULE` 등)을 사용하는 컨테이너를 위한 것이다. 도커 데몬에서 사용자 네임스페이스 재 매핑이 활성화된 경우에만 활성화해야 한다. +- `ExternalPolicyForExternalIP`: ExternalTrafficPolicy가 서비스(Service) ExternalIP에 적용되지 않는 버그를 수정한다. - `GCERegionalPersistentDisk`: GCE에서 지역 PD 기능을 활성화한다. - `GenericEphemeralVolume`: 일반 볼륨의 모든 기능을 지원하는 임시, 인라인 볼륨을 활성화한다(타사 스토리지 공급 업체, 스토리지 용량 추적, 스냅샷으로부터 복원 diff --git a/content/ko/docs/reference/glossary/pod-disruption.md b/content/ko/docs/reference/glossary/pod-disruption.md new file mode 100644 index 0000000000..93a473035c --- /dev/null +++ b/content/ko/docs/reference/glossary/pod-disruption.md @@ -0,0 +1,19 @@ +--- +id: pod-disruption +title: 파드 중단(Disruption) +full_link: /ko/docs/concepts/workloads/pods/disruptions/ +date: 2021-05-12 +short_description: > + 노드에 있는 파드가 자발적 또는 비자발적으로 종료되는 절차 + +aka: +related: + - pod + - container +tags: + - operation +--- + +[파드 중단](/ko/docs/concepts/workloads/pods/disruptions/)은 노드에 있는 파드가 자발적 또는 비자발적으로 종료되는 절차이다. + +자발적 중단은 애플리케이션 소유자 또는 클러스터 관리자가 의도적으로 시작한다. 비자발적 중단은 의도하지 않은 것이며, 노드의 리소스 부족과 같은 피할 수 없는 문제 또는 우발적인 삭제로 인해 트리거될 수 있다. From 70d80a89e650f18341cc48ff4c27b9f11db8935d Mon Sep 17 00:00:00 2001 From: Jihoon Seo Date: Sat, 29 May 2021 22:17:09 +0900 Subject: [PATCH 03/10] [ko] Update outdated files in dev-1.21-ko.3 (p3) --- .../kube-proxy.md | 137 +++++++++++++++++- .../ko/docs/reference/kubectl/cheatsheet.md | 4 + .../ko/docs/reference/scheduling/policies.md | 14 -- .../setup/best-practices/cluster-large.md | 6 +- 4 files changed, 144 insertions(+), 17 deletions(-) diff --git a/content/ko/docs/reference/command-line-tools-reference/kube-proxy.md b/content/ko/docs/reference/command-line-tools-reference/kube-proxy.md index bd930180cd..eab89638db 100644 --- a/content/ko/docs/reference/command-line-tools-reference/kube-proxy.md +++ b/content/ko/docs/reference/command-line-tools-reference/kube-proxy.md @@ -42,6 +42,20 @@ kube-proxy [flags] + +--add-dir-header + + +

true로 되어 있으면, 로그 메시지의 헤더에 파일 디렉터리를 기재한다.

+ + + +--alsologtostderr + + +

로그를 파일뿐만 아니라 표준 에러(standard error)로도 출력한다.

+ + --azure-container-registry-config string @@ -63,6 +77,13 @@ kube-proxy [flags]

true인 경우 kube-proxy는 포트 바인딩 실패를 치명적인 것으로 간주하고 종료한다.

+ +--boot-id-file string     기본값: "/proc/sys/kernel/random/boot_id" + + +

boot-id를 위해 확인할 파일 목록(쉼표로 분리). 가장 먼저 발견되는 항목을 사용한다.

+ + --cleanup @@ -70,6 +91,20 @@ kube-proxy [flags]

true인 경우 iptables 및 ipvs 규칙을 제거하고 종료한다.

+ +--cloud-provider-gce-l7lb-src-cidrs cidrs     기본값: 130.211.0.0/22,35.191.0.0/16 + + +

GCE 방화벽에서, L7 로드밸런싱 트래픽 프록시와 헬스 체크를 위해 개방할 CIDR 목록

+ + + +--cloud-provider-gce-lb-src-cidrs cidrs     기본값: 130.211.0.0/22,209.85.152.0/22,209.85.204.0/22,35.191.0.0/16 + + +

GCE 방화벽에서, L4 로드밸런싱 트래픽 프록시와 헬스 체크를 위해 개방할 CIDR 목록

+ + --cluster-cidr string @@ -119,6 +154,20 @@ kube-proxy [flags]

설정된 TCP 연결에 대한 유휴시간 초과(값이 0이면 그대로 유지)

+ +--default-not-ready-toleration-seconds int     기본값: 300 + + +

notReady:NoExecute 상태에 대한 톨러레이션(toleration) 시간이 지정되지 않은 모든 파드에 기본값으로 지정될 톨러레이션 시간(단위: 초)

+ + + +--default-unreachable-toleration-seconds int     기본값: 300 + + +

unreachable:NoExecute 상태에 대한 톨러레이션 시간이 지정되지 않은 모든 파드에 기본값으로 지정될 톨러레이션 시간(단위: 초)

+ + --detect-local-mode LocalMode @@ -259,6 +308,34 @@ kube-proxy [flags]

인증 정보가 있는 kubeconfig 파일의 경로(마스터 위치는 마스터 플래그로 설정됨).

+ +--log-backtrace-at <'file:N' 형태의 문자열>     기본값: :0 + + +

로깅 과정에서 file:N 번째 라인에 도달하면 스택 트레이스를 출력한다.

+ + + +--log-dir string + + +

로그 파일이 저장될 디렉터리

+ + + +--log-file string + + +

사용할 로그 파일

+ + + +--log-file-max-size uint     기본값: 1800 + + +

로그 파일의 최대 크기(단위: MB). 0으로 설정하면 무제한이다.

+ + --log-flush-frequency duration     기본값: 5s @@ -266,6 +343,20 @@ kube-proxy [flags]

로그 플러시 사이의 최대 시간

+ +--logtostderr     기본값: true + + +

로그를 파일에 기록하지 않고 표준 에러로만 출력

+ + + +--machine-id-file string     기본값: "/etc/machine-id,/var/lib/dbus/machine-id" + + +

machine-id를 위해 확인할 파일 목록(쉼표로 분리). 가장 먼저 발견되는 항목을 사용한다.

+ + --masquerade-all @@ -294,6 +385,13 @@ kube-proxy [flags]

NodePort에 사용할 주소를 지정하는 값의 문자열 조각. 값은 유효한 IP 블록(예: 1.2.3.0/24, 1.2.3.4/32). 기본값인 빈 문자열 조각값은([]) 모든 로컬 주소를 사용하는 것을 의미한다.

+ +--one-output + + +

true이면, 해당 로그가 속하는 심각성 레벨에만 각 로그를 기록한다(원래는 하위 심각성 레벨에도 기록한다).

+ + --oom-score-adj int32     기본값: -999 @@ -326,7 +424,28 @@ kube-proxy [flags] --show-hidden-metrics-for-version string -

숨겨진 메트릭을 표시할 이전 버전.

+

숨겨진 메트릭을 표시할 이전 버전. 이전 마이너 버전만 인식하며, 다른 값은 허용하지 않는다. '1.16' 형태로 사용한다. 이 옵션의 존재 목적은, 다음 릴리스에서 추가적인 메트릭을 숨기는지에 대한 여부를 사용자가 알게 하여, 그 이후 릴리스에서 메트릭이 영구적으로 삭제됐을 때 사용자가 놀라지 않도록 하기 위함이다.

+ + + +--skip-headers + + +

true이면, 로그 메시지에서 헤더 접두사를 붙이지 않는다.

+ + + +--skip-log-headers + + +

true이면, 로그 파일을 열 때 헤더를 붙이지 않는다.

+ + + +--stderrthreshold int     기본값: 2 + + +

이 값 이상의 로그는 표준 에러(stderr)로 출력되도록 한다.

@@ -336,11 +455,25 @@ kube-proxy [flags]

유휴 UDP 연결이 열린 상태로 유지되는 시간(예: '250ms', '2s'). 값은 0보다 커야 한다. 프록시 모드 userspace에만 적용 가능함.

+ +-v, --v int + + +

로그 상세 레벨(verbosity)

+ + --version version[=true] -

버전 정보를 인쇄하고 종료.

+

버전 정보를 출력하고 종료

+ + + +--vmodule <쉼표로 구분된 'pattern=N' 설정> + + +

파일-필터된 로깅을 위한 'pattern=N' 설정들(쉼표로 구분됨)

diff --git a/content/ko/docs/reference/kubectl/cheatsheet.md b/content/ko/docs/reference/kubectl/cheatsheet.md index 4ee9c5406e..71fc5a5f27 100644 --- a/content/ko/docs/reference/kubectl/cheatsheet.md +++ b/content/ko/docs/reference/kubectl/cheatsheet.md @@ -212,6 +212,10 @@ kubectl get nodes -o json | jq -c 'path(..)|[.[]|tostring]|join(".")' # 파드 등에 대해 반환된 모든 키의 마침표로 구분된 트리를 생성한다. kubectl get pods -o json | jq -c 'path(..)|[.[]|tostring]|join(".")' + +# 모든 파드에 대해 ENV를 생성한다(각 파드에 기본 컨테이너가 있고, 기본 네임스페이스가 있고, `env` 명령어가 동작한다고 가정). +# `env` 뿐만 아니라 다른 지원되는 명령어를 모든 파드에 실행할 때에도 참고할 수 있다. +for pod in $(kubectl get po --output=jsonpath={.items..metadata.name}); do echo $pod && kubectl exec -it $pod env; done ``` ## 리소스 업데이트 diff --git a/content/ko/docs/reference/scheduling/policies.md b/content/ko/docs/reference/scheduling/policies.md index 626e077784..c2b9cbbdff 100644 --- a/content/ko/docs/reference/scheduling/policies.md +++ b/content/ko/docs/reference/scheduling/policies.md @@ -37,20 +37,6 @@ weight: 10 - `MaxCSIVolumeCount`: 연결해야 하는 {{< glossary_tooltip text="CSI" term_id="csi" >}} 볼륨의 수와 구성된 제한을 초과하는지 여부를 결정한다. -- `CheckNodeMemoryPressure`: 노드가 메모리 압박을 보고하고 있고, 구성된 - 예외가 없는 경우, 파드가 해당 노드에 스케줄되지 않는다. - -- `CheckNodePIDPressure`: 노드가 프로세스 ID 부족을 보고하고 있고, 구성된 - 예외가 없는 경우, 파드가 해당 노드에 스케줄되지 않는다. - -- `CheckNodeDiskPressure`: 노드가 스토리지 압박(파일시스템이 가득차거나 - 거의 꽉 참)을 보고하고 있고, 구성된 예외가 없는 경우, 파드가 해당 노드에 스케줄되지 않는다. - -- `CheckNodeCondition`: 노드는 파일시스템이 완전히 가득찼거나, - 네트워킹을 사용할 수 없거나, kubelet이 파드를 실행할 준비가 되지 않았다고 보고할 수 있다. - 노드에 대해 이러한 조건이 설정되고, 구성된 예외가 없는 경우, 파드가 - 해당 노드에 스케줄되지 않는다. - - `PodToleratesNodeTaints`: 파드의 {{< glossary_tooltip text="톨러레이션" term_id="toleration" >}}이 노드의 {{< glossary_tooltip text="테인트" term_id="taint" >}}를 용인할 수 있는지 확인한다. diff --git a/content/ko/docs/setup/best-practices/cluster-large.md b/content/ko/docs/setup/best-practices/cluster-large.md index d67892e6dc..d0293e72f6 100644 --- a/content/ko/docs/setup/best-practices/cluster-large.md +++ b/content/ko/docs/setup/best-practices/cluster-large.md @@ -60,9 +60,13 @@ _A_ 영역에 있는 컨트롤 플레인 호스트로만 전달한다. 단일 클러스터 생성시의 부가 스트립트이다. 클러스터 생성 시에 (사용자 도구를 사용하여) 다음을 수행할 수 있다. -* 추가 ectd 인스턴스 시작 및 설정 +* 추가 etcd 인스턴스 시작 및 설정 * 이벤트를 저장하기 위한 {{< glossary_tooltip term_id="kube-apiserver" text="API server" >}} 설정 +[쿠버네티스를 위한 etcd 클러스터 운영하기](/docs/tasks/administer-cluster/configure-upgrade-etcd/)와 +[kubeadm을 이용하여 고가용성 etcd 생성하기](/docs/setup/production-environment/tools/kubeadm/setup-ha-etcd-with-kubeadm/)에서 +큰 클러스터를 위한 etcd를 설정하고 관리하는 방법에 대한 상세 사항을 확인한다. + ## 애드온 리소스 쿠버네티스 [리소스 제한](/ko/docs/concepts/configuration/manage-resources-containers/)은 From 1a3e0e3c139c03354ce049785ad5d540006049da Mon Sep 17 00:00:00 2001 From: seokho-son Date: Sun, 30 May 2021 19:22:07 +0900 Subject: [PATCH 04/10] Restruct release directory and docs for Korean --- content/ko/docs/setup/release/_index.md | 5 - content/ko/docs/setup/release/notes.md | 1626 ----------------- content/ko/releases/_index.md | 27 + content/ko/releases/notes.md | 13 + .../version-skew-policy.md | 4 +- 5 files changed, 42 insertions(+), 1633 deletions(-) delete mode 100755 content/ko/docs/setup/release/_index.md delete mode 100644 content/ko/docs/setup/release/notes.md create mode 100644 content/ko/releases/_index.md create mode 100644 content/ko/releases/notes.md rename content/ko/{docs/setup/release => releases}/version-skew-policy.md (97%) diff --git a/content/ko/docs/setup/release/_index.md b/content/ko/docs/setup/release/_index.md deleted file mode 100755 index fcef7a59ab..0000000000 --- a/content/ko/docs/setup/release/_index.md +++ /dev/null @@ -1,5 +0,0 @@ ---- -title: "릴리스 노트와 버전 차이 지원(skew)" -weight: 10 ---- - diff --git a/content/ko/docs/setup/release/notes.md b/content/ko/docs/setup/release/notes.md deleted file mode 100644 index ada4903931..0000000000 --- a/content/ko/docs/setup/release/notes.md +++ /dev/null @@ -1,1626 +0,0 @@ ---- -title: v1.21 릴리스 노트 -weight: 10 -card: - name: release-notes - weight: 20 - anchors: - - anchor: "#" - title: 현재 릴리스 노트 - - anchor: "#긴급-업그레이드-노트" - title: 긴급 업그레이드 노트 ---- - - - -# v1.21.0 - -[문서](https://docs.k8s.io) - -## v1.21.0 다운로드 - -### 소스 코드 - -파일명 | sha512 해시 --------- | ----------- -[kubernetes.tar.gz](https://dl.k8s.io/v1.21.0/kubernetes.tar.gz) | `19bb76a3fa5ce4b9f043b2a3a77c32365ab1fcb902d8dd6678427fb8be8f49f64a5a03dc46aaef9c7dadee05501cf83412eda46f0edacbb8fc1ed0bf5fb79142` -[kubernetes-src.tar.gz](https://dl.k8s.io/v1.21.0/kubernetes-src.tar.gz) | `f942e6d6c10007a6e9ce21e94df597015ae646a7bc3e515caf1a3b79f1354efb9aff59c40f2553a8e3d43fe4a01742241f5af18b69666244906ed11a22e3bc49` - -### 클라이언트 바이너리 - -파일명 | sha512 해시 --------- | ----------- -[kubernetes-client-darwin-amd64.tar.gz](https://dl.k8s.io/v1.21.0/kubernetes-client-darwin-amd64.tar.gz) | `be9d1440e418e5253fb8a3d8aba705ca8160746a9bd17325ad626a986b6da9f733af864155a651a32b7bca94b533b8d596005ddbe5248bdeea85db47a1b957ed` -[kubernetes-client-darwin-arm64.tar.gz](https://dl.k8s.io/v1.21.0/kubernetes-client-darwin-arm64.tar.gz) | `eed0ddc81d104bb2d41ace13f737c490423d5df4ebddc7376e45c18ed66af35933c9376b912c1c3da105945b04056f6ca0870c156bee8a307cf4189ca5eb1dd1` -[kubernetes-client-linux-386.tar.gz](https://dl.k8s.io/v1.21.0/kubernetes-client-linux-386.tar.gz) | `8a2f30c4434199762f2a96141dab4241c1cce2711bea9ea39cc63c2c5e7d31719ed7f076efac1931604e3a94578d3bbf0cfa454965708c96f3cfb91789868746` -[kubernetes-client-linux-amd64.tar.gz](https://dl.k8s.io/v1.21.0/kubernetes-client-linux-amd64.tar.gz) | `cd3cfa645fa31de3716f1f63506e31b73d2aa8d37bb558bb3b3e8c151f35b3d74d44e03cbd05be67e380f9a5d015aba460222afdac6677815cd99a85c2325cf0` -[kubernetes-client-linux-arm.tar.gz](https://dl.k8s.io/v1.21.0/kubernetes-client-linux-arm.tar.gz) | `936042aa11cea0f6dfd2c30fc5dbe655420b34799bede036b1299a92d6831f589ca10290b73b9c9741560b603ae31e450ad024e273f2b4df5354bfac272691d8` -[kubernetes-client-linux-arm64.tar.gz](https://dl.k8s.io/v1.21.0/kubernetes-client-linux-arm64.tar.gz) | `42beb75364d7bf4bf526804b8a35bd0ab3e124b712e9d1f45c1b914e6be0166619b30695feb24b3eecef134991dacb9ab3597e788bd9e45cf35addddf20dd7f6` -[kubernetes-client-linux-ppc64le.tar.gz](https://dl.k8s.io/v1.21.0/kubernetes-client-linux-ppc64le.tar.gz) | `4baba2ed7046b28370eccc22e2378ae79e3ce58220d6f4f1b6791e8233bec8379e30200bb20b971456b83f2b791ea166fdfcf1ea56908bc1eea03590c0eda468` -[kubernetes-client-linux-s390x.tar.gz](https://dl.k8s.io/v1.21.0/kubernetes-client-linux-s390x.tar.gz) | `37fa0c4d703aef09ce68c10ef3e7362b0313c8f251ce38eea579cd18fae4023d3d2b70e0f31577cabe6958ab9cfc30e98d25a7c64e69048b423057c3cf728339` -[kubernetes-client-windows-386.tar.gz](https://dl.k8s.io/v1.21.0/kubernetes-client-windows-386.tar.gz) | `6900db36c1e3340edfd6dfd8d720575a904c932d39a8a7fa36401595e971a0235bd42111dbcc1cbb77e7374e47f1380a68c637997c18f96a0d9cdc9f3714c4c9` -[kubernetes-client-windows-amd64.tar.gz](https://dl.k8s.io/v1.21.0/kubernetes-client-windows-amd64.tar.gz) | `90de67f6f79fc63bcfdf35066e3d84501cc85433265ffad36fd1a7a428a31b446249f0644a1e97495ea8b2a08e6944df6ef30363003750339edaa2aceffe937c` - -### 서버 바이너리 - -파일명 | sha512 해시 --------- | ----------- -[kubernetes-server-linux-amd64.tar.gz](https://dl.k8s.io/v1.21.0/kubernetes-server-linux-amd64.tar.gz) | `3941dcc2309ac19ec185603a79f5a086d8a198f98c04efa23f15a177e5e1f34946ea9392ba9f5d24d0d727839438f067fef1001fc6e88b27b8b01e35bbd962ca` -[kubernetes-server-linux-arm.tar.gz](https://dl.k8s.io/v1.21.0/kubernetes-server-linux-arm.tar.gz) | `6507abf6c2ec2b336901dc23269f6c577ec0049b8bad3c9dd6ad63f21aa10f09bfbbfa6e064c2466d250411d3e10f8672791a9e10942e38de7bfbaf7a8bcc9da` -[kubernetes-server-linux-arm64.tar.gz](https://dl.k8s.io/v1.21.0/kubernetes-server-linux-arm64.tar.gz) | `5abe76f867ca6865344e957bf166b81766c049ec4eb183a8a5580c22a7f8474db1edf90fd901a5833e56128b6825811653a1d27f72fd34ce5b1287a8c10da05c` -[kubernetes-server-linux-ppc64le.tar.gz](https://dl.k8s.io/v1.21.0/kubernetes-server-linux-ppc64le.tar.gz) | `62507b182ca25396a285d91241536860e58f54fac937e97cbdf91948c83bb41be97d33277400489bf50e85164d560205540b76e94e5d519892312bdc63df1067` -[kubernetes-server-linux-s390x.tar.gz](https://dl.k8s.io/v1.21.0/kubernetes-server-linux-s390x.tar.gz) | `04f2a1f7d1388e4a7d7d9f597f872a3da36f26839cfed16aad6df07021c03f4dca1df06b19cfda56df09d1c2d9a13ebd0af40ca1b9b6aecfaf427ab7712d88f3` - -### 노드 바이너리 - -파일명 | sha512 해시 --------- | ----------- -[kubernetes-node-linux-amd64.tar.gz](https://dl.k8s.io/v1.21.0/kubernetes-node-linux-amd64.tar.gz) | `c1831c708109c31b3878e5a9327ea4b9e546504d0b6b00f3d43db78b5dd7d5114d32ac24a9a505f9cadbe61521f0419933348d2cd309ed8cfe3987d9ca8a7e2c` -[kubernetes-node-linux-arm.tar.gz](https://dl.k8s.io/v1.21.0/kubernetes-node-linux-arm.tar.gz) | `b68dd5bcfc7f9ce2781952df40c8c3a64c29701beff6ac22f042d6f31d4de220e9200b7e8272ddf608114327770acdaf3cb9a34a0a5206e784bda717ea080e0f` -[kubernetes-node-linux-arm64.tar.gz](https://dl.k8s.io/v1.21.0/kubernetes-node-linux-arm64.tar.gz) | `7fa84fc500c28774ed25ca34b6f7b208a2bea29d6e8379f84b9f57bd024aa8fe574418cee7ee26edd55310716d43d65ae7b9cbe11e40c995fe2eac7f66bdb423` -[kubernetes-node-linux-ppc64le.tar.gz](https://dl.k8s.io/v1.21.0/kubernetes-node-linux-ppc64le.tar.gz) | `a4278b3f8e458e9581e01f0c5ba8443303c987988ee136075a8f2f25515d70ca549fbd2e4d10eefca816c75c381d62d71494bd70c47034ab47f8315bbef4ae37` -[kubernetes-node-linux-s390x.tar.gz](https://dl.k8s.io/v1.21.0/kubernetes-node-linux-s390x.tar.gz) | `8de2bc6f22f232ff534b45012986eac23893581ccb6c45bd637e40dbe808ce31d5a92375c00dc578bdbadec342b6e5b70c1b9f3d3a7bb26ccfde97d71f9bf84a` -[kubernetes-node-windows-amd64.tar.gz](https://dl.k8s.io/v1.21.0/kubernetes-node-windows-amd64.tar.gz) | `b82e94663d330cff7a117f99a7544f27d0bc92b36b5a283b3c23725d5b33e6f15e0ebf784627638f22f2d58c58c0c2b618ddfd226a64ae779693a0861475d355` - -## v1.20.0 이후 변경로그 (Changelog) - -## 새로운 소식 (주요 테마) - -### Deprecation of PodSecurityPolicy - -PSP as an admission controller resource is being deprecated. Deployed PodSecurityPolicy's will keep working until version 1.25, their target removal from the codebase. A new feature, with a working title of "PSP replacement policy", is being developed in [KEP-2579](https://features.k8s.io/2579). To learn more, read [PodSecurityPolicy Deprecation: Past, Present, and Future](https://blog.k8s.io/2021/04/06/podsecuritypolicy-deprecation-past-present-and-future/). - -### Kubernetes API Reference Documentation - -The API reference is now generated with [`gen-resourcesdocs`](https://github.com/kubernetes-sigs/reference-docs/tree/c96658d89fb21037b7d00d27e6dbbe6b32375837/gen-resourcesdocs) and it is moving to [Kubernetes API](https://docs.k8s.io/reference/kubernetes-api/) - -### Kustomize Updates in Kubectl - -[Kustomize](https://github.com/kubernetes-sigs/kustomize) version in kubectl had a jump from v2.0.3 to [v4.0.5](https://github.com/kubernetes/kubernetes/pull/98946). Kustomize is now treated as a library and future updates will be less sporadic. - -### Default Container Labels - -Pod with multiple containers can use `kubectl.kubernetes.io/default-container` label to have a container preselected for kubectl commands. More can be read in [KEP-2227](https://github.com/kubernetes/enhancements/blob/master/keps/sig-cli/2227-kubectl-default-container/README.md). - -### Immutable Secrets and ConfigMaps - -Immutable Secrets and ConfigMaps graduates to GA. This feature allows users to specify that the contents of a particular Secret or ConfigMap is immutable for its object lifetime. For such instances, Kubelet will not watch/poll for changes and therefore reducing apiserver load. - -### Structured Logging in Kubelet - -Kubelet has adopted structured logging, thanks to community effort in accomplishing this within the release timeline. Structured logging in the project remains an ongoing effort -- for folks interested in participating, [keep an eye / chime in to the mailing list discussion](https://groups.google.com/g/kubernetes-dev/c/y4WIw-ntUR8). - -### Storage Capacity Tracking - -Traditionally, the Kubernetes scheduler was based on the assumptions that additional persistent storage is available everywhere in the cluster and has infinite capacity. Topology constraints addressed the first point, but up to now pod scheduling was still done without considering that the remaining storage capacity may not be enough to start a new pod. [Storage capacity tracking](https://docs.k8s.io/concepts/storage/storage-capacity/) addresses that by adding an API for a CSI driver to report storage capacity and uses that information in the Kubernetes scheduler when choosing a node for a pod. This feature serves as a stepping stone for supporting dynamic provisioning for local volumes and other volume types that are more capacity constrained. - -### Generic Ephemeral Volumes - -[Generic ephermeral volumes](https://docs.k8s.io/concepts/storage/ephemeral-volumes/#generic-ephemeral-volumes) feature allows any existing storage driver that supports dynamic provisioning to be used as an ephemeral volume with the volume’s lifecycle bound to the Pod. It can be used to provide scratch storage that is different from the root disk, for example persistent memory, or a separate local disk on that node. All StorageClass parameters for volume provisioning are supported. All features supported with PersistentVolumeClaims are supported, such as storage capacity tracking, snapshots and restore, and volume resizing. - -### CSI Service Account Token - -CSI Service Account Token feature moves to Beta in 1.21. This feature improves the security posture and allows CSI drivers to receive pods' [bound service account tokens](https://github.com/kubernetes/enhancements/blob/master/keps/sig-auth/1205-bound-service-account-tokens/README.md). This feature also provides a knob to re-publish volumes so that short-lived volumes can be refreshed. - -### CSI Health Monitoring - -The CSI health monitoring feature is being released as a second Alpha in Kubernetes 1.21. This feature enables CSI Drivers to share abnormal volume conditions from the underlying storage systems with Kubernetes so that they can be reported as events on PVCs or Pods. This feature serves as a stepping stone towards programmatic detection and resolution of individual volume health issues by Kubernetes. - -## 알려진 이슈 - -### `TopologyAwareHints` feature falls back to default behavior - -The feature gate currently falls back to the default behavior in most cases. Enabling the feature gate will add hints to `EndpointSlices`, but functional differences are only observed in non-dual stack kube-proxy implementation. [The fix will be available in coming releases](https://github.com/kubernetes/kubernetes/pull/100804). - -## 긴급 업그레이드 노트 - -### (주의. 업그레이드 전에 반드시 읽어야 함) - -- Kube-proxy's IPVS proxy mode no longer sets the net.ipv4.conf.all.route_localnet sysctl parameter. Nodes upgrading will have net.ipv4.conf.all.route_localnet set to 1 but new nodes will inherit the system default (usually 0). If you relied on any behavior requiring net.ipv4.conf.all.route_localnet, you must set ensure it is enabled as kube-proxy will no longer set it automatically. This change helps to further mitigate CVE-2020-8558. ([#92938](https://github.com/kubernetes/kubernetes/pull/92938), [@lbernail](https://github.com/lbernail)) [SIG Network and Release] - - Kubeadm: during "init" an empty cgroupDriver value in the KubeletConfiguration is now always set to "systemd" unless the user is explicit about it. This requires existing machine setups to configure the container runtime to use the "systemd" driver. Documentation on this topic can be found here: https://kubernetes.io/docs/setup/production-environment/container-runtimes/. When upgrading existing clusters / nodes using "kubeadm upgrade" the old cgroupDriver value is preserved, but in 1.22 this change will also apply to "upgrade". For more information on migrating to the "systemd" driver or remaining on the "cgroupfs" driver see: https://kubernetes.io/docs/tasks/administer-cluster/kubeadm/configure-cgroup-driver/. ([#99471](https://github.com/kubernetes/kubernetes/pull/99471), [@neolit123](https://github.com/neolit123)) [SIG Cluster Lifecycle] - - Newly provisioned PVs by EBS plugin will no longer use the deprecated "failure-domain.beta.kubernetes.io/zone" and "failure-domain.beta.kubernetes.io/region" labels. It will use "topology.kubernetes.io/zone" and "topology.kubernetes.io/region" labels instead. ([#99130](https://github.com/kubernetes/kubernetes/pull/99130), [@ayberk](https://github.com/ayberk)) [SIG Cloud Provider, Storage and Testing] - - Newly provisioned PVs by OpenStack Cinder plugin will no longer use the deprecated "failure-domain.beta.kubernetes.io/zone" and "failure-domain.beta.kubernetes.io/region" labels. It will use "topology.kubernetes.io/zone" and "topology.kubernetes.io/region" labels instead. ([#99719](https://github.com/kubernetes/kubernetes/pull/99719), [@jsafrane](https://github.com/jsafrane)) [SIG Cloud Provider and Storage] - - Newly provisioned PVs by gce-pd will no longer have the beta FailureDomain label. gce-pd volume plugin will start to have GA topology label instead. ([#98700](https://github.com/kubernetes/kubernetes/pull/98700), [@Jiawei0227](https://github.com/Jiawei0227)) [SIG Cloud Provider, Storage and Testing] - - OpenStack Cinder CSI migration is on by default, Clinder CSI driver must be installed on clusters on OpenStack for Cinder volumes to work. ([#98538](https://github.com/kubernetes/kubernetes/pull/98538), [@dims](https://github.com/dims)) [SIG Storage] - - Remove alpha `CSIMigrationXXComplete` flag and add alpha `InTreePluginXXUnregister` flag. Deprecate `CSIMigrationvSphereComplete` flag and it will be removed in v1.22. ([#98243](https://github.com/kubernetes/kubernetes/pull/98243), [@Jiawei0227](https://github.com/Jiawei0227)) - - Remove storage metrics `storage_operation_errors_total`, since we already have `storage_operation_status_count`.And add new field `status` for `storage_operation_duration_seconds`, so that we can know about all status storage operation latency. ([#98332](https://github.com/kubernetes/kubernetes/pull/98332), [@JornShen](https://github.com/JornShen)) [SIG Instrumentation and Storage] - - The metric `storage_operation_errors_total` is not removed, but is marked deprecated, and the metric `storage_operation_status_count` is marked deprecated. In both cases the `storage_operation_duration_seconds` metric can be used to recover equivalent counts (using `status=fail-unknown` in the case of `storage_operations_errors_total`). ([#99045](https://github.com/kubernetes/kubernetes/pull/99045), [@mattcary](https://github.com/mattcary)) - - `ServiceNodeExclusion`, `NodeDisruptionExclusion` and `LegacyNodeRoleBehavior` features have been promoted to GA. `ServiceNodeExclusion` and `NodeDisruptionExclusion` are now unconditionally enabled, while `LegacyNodeRoleBehavior` is unconditionally disabled. To prevent control plane nodes from being added to load balancers automatically, upgrade users need to add "node.kubernetes.io/exclude-from-external-load-balancers" label to control plane nodes. ([#97543](https://github.com/kubernetes/kubernetes/pull/97543), [@pacoxu](https://github.com/pacoxu)) - -## 종류(Kind)별 변경 사항 - -### 사용 중단 - -- Aborting the drain command in a list of nodes will be deprecated. The new behavior will make the drain command go through all nodes even if one or more nodes failed during the drain. For now, users can try such experience by enabling --ignore-errors flag. ([#98203](https://github.com/kubernetes/kubernetes/pull/98203), [@yuzhiquan](https://github.com/yuzhiquan)) -- Delete deprecated `service.beta.kubernetes.io/azure-load-balancer-mixed-protocols` mixed procotol annotation in favor of the MixedProtocolLBService feature ([#97096](https://github.com/kubernetes/kubernetes/pull/97096), [@nilo19](https://github.com/nilo19)) [SIG Cloud Provider] -- Deprecate the `topologyKeys` field in Service. This capability will be replaced with upcoming work around Topology Aware Subsetting and Service Internal Traffic Policy. ([#96736](https://github.com/kubernetes/kubernetes/pull/96736), [@andrewsykim](https://github.com/andrewsykim)) [SIG Apps] -- Kube-proxy: remove deprecated --cleanup-ipvs flag of kube-proxy, and make --cleanup flag always to flush IPVS ([#97336](https://github.com/kubernetes/kubernetes/pull/97336), [@maaoBit](https://github.com/maaoBit)) [SIG Network] -- Kubeadm: deprecated command "alpha selfhosting pivot" is now removed. ([#97627](https://github.com/kubernetes/kubernetes/pull/97627), [@knight42](https://github.com/knight42)) -- Kubeadm: graduate the command `kubeadm alpha kubeconfig user` to `kubeadm kubeconfig user`. The `kubeadm alpha kubeconfig user` command is deprecated now. ([#97583](https://github.com/kubernetes/kubernetes/pull/97583), [@knight42](https://github.com/knight42)) [SIG Cluster Lifecycle] -- Kubeadm: the "kubeadm alpha certs" command is removed now, please use "kubeadm certs" instead. ([#97706](https://github.com/kubernetes/kubernetes/pull/97706), [@knight42](https://github.com/knight42)) [SIG Cluster Lifecycle] -- Kubeadm: the deprecated kube-dns is no longer supported as an option. If "ClusterConfiguration.dns.type" is set to "kube-dns" kubeadm will now throw an error. ([#99646](https://github.com/kubernetes/kubernetes/pull/99646), [@rajansandeep](https://github.com/rajansandeep)) [SIG Cluster Lifecycle] -- Kubectl: The deprecated `kubectl alpha debug` command is removed. Use `kubectl debug` instead. ([#98111](https://github.com/kubernetes/kubernetes/pull/98111), [@pandaamanda](https://github.com/pandaamanda)) [SIG CLI] -- Official support to build kubernetes with docker-machine / remote docker is removed. This change does not affect building kubernetes with docker locally. ([#97935](https://github.com/kubernetes/kubernetes/pull/97935), [@adeniyistephen](https://github.com/adeniyistephen)) [SIG Release and Testing] -- Remove deprecated `--generator, --replicas, --service-generator, --service-overrides, --schedule` from `kubectl run` - Deprecate `--serviceaccount, --hostport, --requests, --limits` in `kubectl run` ([#99732](https://github.com/kubernetes/kubernetes/pull/99732), [@soltysh](https://github.com/soltysh)) -- Remove the deprecated metrics "scheduling_algorithm_preemption_evaluation_seconds" and "binding_duration_seconds", suggest to use "scheduler_framework_extension_point_duration_seconds" instead. ([#96447](https://github.com/kubernetes/kubernetes/pull/96447), [@chendave](https://github.com/chendave)) [SIG Cluster Lifecycle, Instrumentation, Scheduling and Testing] -- Removing experimental windows container hyper-v support with Docker ([#97141](https://github.com/kubernetes/kubernetes/pull/97141), [@wawa0210](https://github.com/wawa0210)) [SIG Node and Windows] -- Rename metrics `etcd_object_counts` to `apiserver_storage_object_counts` and mark it as stable. The original `etcd_object_counts` metrics name is marked as "Deprecated" and will be removed in the future. ([#99785](https://github.com/kubernetes/kubernetes/pull/99785), [@erain](https://github.com/erain)) [SIG API Machinery, Instrumentation and Testing] -- The GA TokenRequest and TokenRequestProjection feature gates have been removed and are unconditionally enabled. Remove explicit use of those feature gates in CLI invocations. ([#97148](https://github.com/kubernetes/kubernetes/pull/97148), [@wawa0210](https://github.com/wawa0210)) [SIG Node] -- The PodSecurityPolicy API is deprecated in 1.21, and will no longer be served starting in 1.25. ([#97171](https://github.com/kubernetes/kubernetes/pull/97171), [@deads2k](https://github.com/deads2k)) [SIG Auth and CLI] -- The `batch/v2alpha1` CronJob type definitions and clients are deprecated and removed. ([#96987](https://github.com/kubernetes/kubernetes/pull/96987), [@soltysh](https://github.com/soltysh)) [SIG API Machinery, Apps, CLI and Testing] -- The `export` query parameter (inconsistently supported by API resources and deprecated in v1.14) is fully removed. Requests setting this query parameter will now receive a 400 status response. ([#98312](https://github.com/kubernetes/kubernetes/pull/98312), [@deads2k](https://github.com/deads2k)) [SIG API Machinery, Auth and Testing] -- `audit.k8s.io/v1beta1` and `audit.k8s.io/v1alpha1` audit policy configuration and audit events are deprecated in favor of `audit.k8s.io/v1`, available since v1.13. kube-apiserver invocations that specify alpha or beta policy configurations with `--audit-policy-file`, or explicitly request alpha or beta audit events with `--audit-log-version` / `--audit-webhook-version` must update to use `audit.k8s.io/v1` and accept `audit.k8s.io/v1` events prior to v1.24. ([#98858](https://github.com/kubernetes/kubernetes/pull/98858), [@carlory](https://github.com/carlory)) [SIG Auth] -- `discovery.k8s.io/v1beta1` EndpointSlices are deprecated in favor of `discovery.k8s.io/v1`, and will no longer be served in Kubernetes v1.25. ([#100472](https://github.com/kubernetes/kubernetes/pull/100472), [@liggitt](https://github.com/liggitt)) -- `diskformat` storage class parameter for in-tree vSphere volume plugin is deprecated as of v1.21 release. Please consider updating storageclass and remove `diskformat` parameter. vSphere CSI Driver does not support diskformat storageclass parameter. - - vSphere releases less than 67u3 are deprecated as of v1.21. Please consider upgrading vSphere to 67u3 or above. vSphere CSI Driver requires minimum vSphere 67u3. - - VM Hardware version less than 15 is deprecated as of v1.21. Please consider upgrading the Node VM Hardware version to 15 or above. vSphere CSI Driver recommends Node VM's Hardware version set to at least vmx-15. - - Multi vCenter support is deprecated as of v1.21. If you have a Kubernetes cluster spanning across multiple vCenter servers, please consider moving all k8s nodes to a single vCenter Server. vSphere CSI Driver does not support Kubernetes deployment spanning across multiple vCenter servers. - - Support for these deprecations will be available till Kubernetes v1.24. ([#98546](https://github.com/kubernetes/kubernetes/pull/98546), [@divyenpatel](https://github.com/divyenpatel)) - -### API 변경 - -- 1. PodAffinityTerm includes a namespaceSelector field to allow selecting eligible namespaces based on their labels. - 2. A new CrossNamespacePodAffinity quota scope API that allows restricting which namespaces allowed to use PodAffinityTerm with corss-namespace reference via namespaceSelector or namespaces fields. ([#98582](https://github.com/kubernetes/kubernetes/pull/98582), [@ahg-g](https://github.com/ahg-g)) [SIG API Machinery, Apps, Auth and Testing] -- Add Probe-level terminationGracePeriodSeconds field ([#99375](https://github.com/kubernetes/kubernetes/pull/99375), [@ehashman](https://github.com/ehashman)) [SIG API Machinery, Apps, Node and Testing] -- Added `.spec.completionMode` field to Job, with accepted values `NonIndexed` (default) and `Indexed`. This is an alpha field and is only honored by servers with the `IndexedJob` feature gate enabled. ([#98441](https://github.com/kubernetes/kubernetes/pull/98441), [@alculquicondor](https://github.com/alculquicondor)) [SIG Apps and CLI] -- Adds support for endPort field in NetworkPolicy ([#97058](https://github.com/kubernetes/kubernetes/pull/97058), [@rikatz](https://github.com/rikatz)) [SIG Apps and Network] -- CSIServiceAccountToken graduates to Beta and enabled by default. ([#99298](https://github.com/kubernetes/kubernetes/pull/99298), [@zshihang](https://github.com/zshihang)) -- Cluster admins can now turn off `/debug/pprof` and `/debug/flags/v` endpoint in kubelet by setting `enableProfilingHandler` and `enableDebugFlagsHandler` to `false` in the Kubelet configuration file. Options `enableProfilingHandler` and `enableDebugFlagsHandler` can be set to `true` only when `enableDebuggingHandlers` is also set to `true`. ([#98458](https://github.com/kubernetes/kubernetes/pull/98458), [@SaranBalaji90](https://github.com/SaranBalaji90)) -- DaemonSets accept a MaxSurge integer or percent on their rolling update strategy that will launch the updated pod on nodes and wait for those pods to go ready before marking the old out-of-date pods as deleted. This allows workloads to avoid downtime during upgrades when deployed using DaemonSets. This feature is alpha and is behind the DaemonSetUpdateSurge feature gate. ([#96441](https://github.com/kubernetes/kubernetes/pull/96441), [@smarterclayton](https://github.com/smarterclayton)) [SIG Apps and Testing] -- Enable SPDY pings to keep connections alive, so that `kubectl exec` and `kubectl portforward` won't be interrupted. ([#97083](https://github.com/kubernetes/kubernetes/pull/97083), [@knight42](https://github.com/knight42)) [SIG API Machinery and CLI] -- FieldManager no longer owns fields that get reset before the object is persisted (e.g. "status wiping"). ([#99661](https://github.com/kubernetes/kubernetes/pull/99661), [@kevindelgado](https://github.com/kevindelgado)) [SIG API Machinery, Auth and Testing] -- Fixes server-side apply for APIService resources. ([#98576](https://github.com/kubernetes/kubernetes/pull/98576), [@kevindelgado](https://github.com/kevindelgado)) -- Generic ephemeral volumes are beta. ([#99643](https://github.com/kubernetes/kubernetes/pull/99643), [@pohly](https://github.com/pohly)) [SIG API Machinery, Apps, Auth, CLI, Node, Storage and Testing] -- Hugepages request values are limited to integer multiples of the page size. ([#98515](https://github.com/kubernetes/kubernetes/pull/98515), [@lala123912](https://github.com/lala123912)) [SIG Apps] -- Implement the GetAvailableResources in the podresources API. ([#95734](https://github.com/kubernetes/kubernetes/pull/95734), [@fromanirh](https://github.com/fromanirh)) [SIG Instrumentation, Node and Testing] -- IngressClass resource can now reference a resource in a specific namespace - for implementation-specific configuration (previously only Cluster-level resources were allowed). - This feature can be enabled using the IngressClassNamespacedParams feature gate. ([#99275](https://github.com/kubernetes/kubernetes/pull/99275), [@hbagdi](https://github.com/hbagdi)) -- Jobs API has a new `.spec.suspend` field that can be used to suspend and resume Jobs. This is an alpha field which is only honored by servers with the `SuspendJob` feature gate enabled. ([#98727](https://github.com/kubernetes/kubernetes/pull/98727), [@adtac](https://github.com/adtac)) -- Kubelet Graceful Node Shutdown feature graduates to Beta and enabled by default. ([#99735](https://github.com/kubernetes/kubernetes/pull/99735), [@bobbypage](https://github.com/bobbypage)) -- Kubernetes is now built using go1.15.7 ([#98363](https://github.com/kubernetes/kubernetes/pull/98363), [@cpanato](https://github.com/cpanato)) [SIG Cloud Provider, Instrumentation, Node, Release and Testing] -- Namespace API objects now have a `kubernetes.io/metadata.name` label matching their metadata.name field to allow selecting any namespace by its name using a label selector. ([#96968](https://github.com/kubernetes/kubernetes/pull/96968), [@jayunit100](https://github.com/jayunit100)) [SIG API Machinery, Apps, Cloud Provider, Storage and Testing] -- One new field "InternalTrafficPolicy" in Service is added. - It specifies if the cluster internal traffic should be routed to all endpoints or node-local endpoints only. - "Cluster" routes internal traffic to a Service to all endpoints. - "Local" routes traffic to node-local endpoints only, and traffic is dropped if no node-local endpoints are ready. - The default value is "Cluster". ([#96600](https://github.com/kubernetes/kubernetes/pull/96600), [@maplain](https://github.com/maplain)) [SIG API Machinery, Apps and Network] -- PodDisruptionBudget API objects can now contain conditions in status. ([#98127](https://github.com/kubernetes/kubernetes/pull/98127), [@mortent](https://github.com/mortent)) [SIG API Machinery, Apps, Auth, CLI, Cloud Provider, Cluster Lifecycle and Instrumentation] -- PodSecurityPolicy only stores "generic" as allowed volume type if the GenericEphemeralVolume feature gate is enabled ([#98918](https://github.com/kubernetes/kubernetes/pull/98918), [@pohly](https://github.com/pohly)) [SIG Auth and Security] -- Promote CronJobs to batch/v1 ([#99423](https://github.com/kubernetes/kubernetes/pull/99423), [@soltysh](https://github.com/soltysh)) [SIG API Machinery, Apps, CLI and Testing] -- Promote Immutable Secrets/ConfigMaps feature to Stable. This allows to set `immutable` field in Secret or ConfigMap object to mark their contents as immutable. ([#97615](https://github.com/kubernetes/kubernetes/pull/97615), [@wojtek-t](https://github.com/wojtek-t)) [SIG Apps, Architecture, Node and Testing] -- Remove support for building Kubernetes with bazel. ([#99561](https://github.com/kubernetes/kubernetes/pull/99561), [@BenTheElder](https://github.com/BenTheElder)) [SIG API Machinery, Apps, Architecture, Auth, Autoscaling, CLI, Cloud Provider, Cluster Lifecycle, Instrumentation, Network, Node, Release, Scalability, Scheduling, Storage, Testing and Windows] -- Scheduler extender filter interface now can report unresolvable failed nodes in the new field `FailedAndUnresolvableNodes` of `ExtenderFilterResult` struct. Nodes in this map will be skipped in the preemption phase. ([#92866](https://github.com/kubernetes/kubernetes/pull/92866), [@cofyc](https://github.com/cofyc)) [SIG Scheduling] -- Services can specify loadBalancerClass to use a custom load balancer ([#98277](https://github.com/kubernetes/kubernetes/pull/98277), [@XudongLiuHarold](https://github.com/XudongLiuHarold)) -- Storage capacity tracking (= the CSIStorageCapacity feature) graduates to Beta and enabled by default, storage.k8s.io/v1alpha1/VolumeAttachment and storage.k8s.io/v1alpha1/CSIStorageCapacity objects are deprecated ([#99641](https://github.com/kubernetes/kubernetes/pull/99641), [@pohly](https://github.com/pohly)) -- Support for Indexed Job: a Job that is considered completed when Pods associated to indexes from 0 to (.spec.completions-1) have succeeded. ([#98812](https://github.com/kubernetes/kubernetes/pull/98812), [@alculquicondor](https://github.com/alculquicondor)) [SIG Apps and CLI] -- The BoundServiceAccountTokenVolume feature has been promoted to beta, and enabled by default. - - This changes the tokens provided to containers at `/var/run/secrets/kubernetes.io/serviceaccount/token` to be time-limited, auto-refreshed, and invalidated when the containing pod is deleted. - - Clients should reload the token from disk periodically (once per minute is recommended) to ensure they continue to use a valid token. `k8s.io/client-go` version v11.0.0+ and v0.15.0+ reload tokens automatically. - - By default, injected tokens are given an extended lifetime so they remain valid even after a new refreshed token is provided. The metric `serviceaccount_stale_tokens_total` can be used to monitor for workloads that are depending on the extended lifetime and are continuing to use tokens even after a refreshed token is provided to the container. If that metric indicates no existing workloads are depending on extended lifetimes, injected token lifetime can be shortened to 1 hour by starting `kube-apiserver` with `--service-account-extend-token-expiration=false`. ([#95667](https://github.com/kubernetes/kubernetes/pull/95667), [@zshihang](https://github.com/zshihang)) [SIG API Machinery, Auth, Cluster Lifecycle and Testing] -- The EndpointSlice Controllers are now GA. The `EndpointSliceController` will not populate the `deprecatedTopology` field and will only provide topology information through the `zone` and `nodeName` fields. ([#99870](https://github.com/kubernetes/kubernetes/pull/99870), [@swetharepakula](https://github.com/swetharepakula)) -- The Endpoints controller will now set the `endpoints.kubernetes.io/over-capacity` annotation to "warning" when an Endpoints resource contains more than 1000 addresses. In a future release, the controller will truncate Endpoints that exceed this limit. The EndpointSlice API can be used to support significantly larger number of addresses. ([#99975](https://github.com/kubernetes/kubernetes/pull/99975), [@robscott](https://github.com/robscott)) [SIG Apps and Network] -- The PodDisruptionBudget API has been promoted to policy/v1 with no schema changes. The only functional change is that an empty selector (`{}`) written to a policy/v1 PodDisruptionBudget now selects all pods in the namespace. The behavior of the policy/v1beta1 API remains unchanged. The policy/v1beta1 PodDisruptionBudget API is deprecated and will no longer be served in 1.25+. ([#99290](https://github.com/kubernetes/kubernetes/pull/99290), [@mortent](https://github.com/mortent)) [SIG API Machinery, Apps, Auth, Autoscaling, CLI, Cloud Provider, Cluster Lifecycle, Instrumentation, Scheduling and Testing] -- The `EndpointSlice` API is now GA. The `EndpointSlice` topology field has been removed from the GA API and will be replaced by a new per Endpoint Zone field. If the topology field was previously used, it will be converted into an annotation in the v1 Resource. The `discovery.k8s.io/v1alpha1` API is removed. ([#99662](https://github.com/kubernetes/kubernetes/pull/99662), [@swetharepakula](https://github.com/swetharepakula)) -- The `controller.kubernetes.io/pod-deletion-cost` annotation can be set to offer a hint on the cost of deleting a `Pod` compared to other pods belonging to the same ReplicaSet. Pods with lower deletion cost are deleted first. This is an alpha feature. ([#99163](https://github.com/kubernetes/kubernetes/pull/99163), [@ahg-g](https://github.com/ahg-g)) -- The kube-apiserver now resets `managedFields` that got corrupted by a mutating admission controller. ([#98074](https://github.com/kubernetes/kubernetes/pull/98074), [@kwiesmueller](https://github.com/kwiesmueller)) -- Topology Aware Hints are now available in alpha and can be enabled with the `TopologyAwareHints` feature gate. ([#99522](https://github.com/kubernetes/kubernetes/pull/99522), [@robscott](https://github.com/robscott)) [SIG API Machinery, Apps, Auth, Instrumentation, Network and Testing] -- Users might specify the `kubectl.kubernetes.io/default-exec-container` annotation in a Pod to preselect container for kubectl commands. ([#97099](https://github.com/kubernetes/kubernetes/pull/97099), [@pacoxu](https://github.com/pacoxu)) [SIG CLI] - -### 기능 (Feature) - -- A client-go metric, rest_client_exec_plugin_call_total, has been added to track total calls to client-go credential plugins. ([#98892](https://github.com/kubernetes/kubernetes/pull/98892), [@ankeesler](https://github.com/ankeesler)) [SIG API Machinery, Auth, Cluster Lifecycle and Instrumentation] -- A new histogram metric to track the time it took to delete a job by the `TTLAfterFinished` controller ([#98676](https://github.com/kubernetes/kubernetes/pull/98676), [@ahg-g](https://github.com/ahg-g)) -- AWS cloud provider supports auto-discovering subnets without any `kubernetes.io/cluster/` tags. It also supports additional service annotation `service.beta.kubernetes.io/aws-load-balancer-subnets` to manually configure the subnets. ([#97431](https://github.com/kubernetes/kubernetes/pull/97431), [@kishorj](https://github.com/kishorj)) -- Aborting the drain command in a list of nodes will be deprecated. The new behavior will make the drain command go through all nodes even if one or more nodes failed during the drain. For now, users can try such experience by enabling --ignore-errors flag. ([#98203](https://github.com/kubernetes/kubernetes/pull/98203), [@yuzhiquan](https://github.com/yuzhiquan)) -- Add --permit-address-sharing flag to `kube-apiserver` to listen with `SO_REUSEADDR`. While allowing to listen on wildcard IPs like 0.0.0.0 and specific IPs in parallel, it avoids waiting for the kernel to release socket in `TIME_WAIT` state, and hence, considerably reducing `kube-apiserver` restart times under certain conditions. ([#93861](https://github.com/kubernetes/kubernetes/pull/93861), [@sttts](https://github.com/sttts)) -- Add `csi_operations_seconds` metric on kubelet that exposes CSI operations duration and status for node CSI operations. ([#98979](https://github.com/kubernetes/kubernetes/pull/98979), [@Jiawei0227](https://github.com/Jiawei0227)) [SIG Instrumentation and Storage] -- Add `migrated` field into `storage_operation_duration_seconds` metric ([#99050](https://github.com/kubernetes/kubernetes/pull/99050), [@Jiawei0227](https://github.com/Jiawei0227)) [SIG Apps, Instrumentation and Storage] -- Add flag --lease-reuse-duration-seconds for kube-apiserver to config etcd lease reuse duration. ([#97009](https://github.com/kubernetes/kubernetes/pull/97009), [@lingsamuel](https://github.com/lingsamuel)) [SIG API Machinery and Scalability] -- Add metric etcd_lease_object_counts for kube-apiserver to observe max objects attached to a single etcd lease. ([#97480](https://github.com/kubernetes/kubernetes/pull/97480), [@lingsamuel](https://github.com/lingsamuel)) [SIG API Machinery, Instrumentation and Scalability] -- Add support to generate client-side binaries for new darwin/arm64 platform ([#97743](https://github.com/kubernetes/kubernetes/pull/97743), [@dims](https://github.com/dims)) [SIG Release and Testing] -- Added `ephemeral_volume_controller_create[_failures]_total` counters to kube-controller-manager metrics ([#99115](https://github.com/kubernetes/kubernetes/pull/99115), [@pohly](https://github.com/pohly)) [SIG API Machinery, Apps, Cluster Lifecycle, Instrumentation and Storage] -- Added support for installing `arm64` node artifacts. ([#99242](https://github.com/kubernetes/kubernetes/pull/99242), [@liu-cong](https://github.com/liu-cong)) -- Adds alpha feature `VolumeCapacityPriority` which makes the scheduler prioritize nodes based on the best matching size of statically provisioned PVs across multiple topologies. ([#96347](https://github.com/kubernetes/kubernetes/pull/96347), [@cofyc](https://github.com/cofyc)) [SIG Apps, Network, Scheduling, Storage and Testing] -- Adds the ability to pass --strict-transport-security-directives to the kube-apiserver to set the HSTS header appropriately. Be sure you understand the consequences to browsers before setting this field. ([#96502](https://github.com/kubernetes/kubernetes/pull/96502), [@249043822](https://github.com/249043822)) [SIG Auth] -- Adds two new metrics to cronjobs, a histogram to track the time difference when a job is created and the expected time when it should be created, as well as a gauge for the missed schedules of a cronjob ([#99341](https://github.com/kubernetes/kubernetes/pull/99341), [@alaypatel07](https://github.com/alaypatel07)) -- Alpha implementation of Kubectl Command Headers: SIG CLI KEP 859 enabled when KUBECTL_COMMAND_HEADERS environment variable set on the client command line. ([#98952](https://github.com/kubernetes/kubernetes/pull/98952), [@seans3](https://github.com/seans3)) -- Base-images: Update to debian-iptables:buster-v1.4.0 - - Uses iptables 1.8.5 - - base-images: Update to debian-base:buster-v1.3.0 - - cluster/images/etcd: Build etcd:3.4.13-2 image - - Uses debian-base:buster-v1.3.0 ([#98401](https://github.com/kubernetes/kubernetes/pull/98401), [@pacoxu](https://github.com/pacoxu)) [SIG Testing] -- CRIContainerLogRotation graduates to GA and unconditionally enabled. ([#99651](https://github.com/kubernetes/kubernetes/pull/99651), [@umohnani8](https://github.com/umohnani8)) -- Component owner can configure the allowlist of metric label with flag '--allow-metric-labels'. ([#99385](https://github.com/kubernetes/kubernetes/pull/99385), [@YoyinZyc](https://github.com/YoyinZyc)) [SIG API Machinery, CLI, Cloud Provider, Cluster Lifecycle, Instrumentation and Release] -- Component owner can configure the allowlist of metric label with flag '--allow-metric-labels'. ([#99738](https://github.com/kubernetes/kubernetes/pull/99738), [@YoyinZyc](https://github.com/YoyinZyc)) [SIG API Machinery, Cluster Lifecycle and Instrumentation] -- EmptyDir memory backed volumes are sized as the the minimum of pod allocatable memory on a host and an optional explicit user provided value. ([#100319](https://github.com/kubernetes/kubernetes/pull/100319), [@derekwaynecarr](https://github.com/derekwaynecarr)) [SIG Node] -- Enables Kubelet to check volume condition and log events to corresponding pods. ([#99284](https://github.com/kubernetes/kubernetes/pull/99284), [@fengzixu](https://github.com/fengzixu)) [SIG Apps, Instrumentation, Node and Storage] -- EndpointSliceNodeName graduates to GA and thus will be unconditionally enabled -- NodeName will always be available in the v1beta1 API. ([#99746](https://github.com/kubernetes/kubernetes/pull/99746), [@swetharepakula](https://github.com/swetharepakula)) -- Export `NewDebuggingRoundTripper` function and `DebugLevel` options in the k8s.io/client-go/transport package. ([#98324](https://github.com/kubernetes/kubernetes/pull/98324), [@atosatto](https://github.com/atosatto)) -- Kube-proxy iptables: new metric sync_proxy_rules_iptables_total that exposes the number of rules programmed per table in each iteration ([#99653](https://github.com/kubernetes/kubernetes/pull/99653), [@aojea](https://github.com/aojea)) [SIG Instrumentation and Network] -- Kube-scheduler now logs plugin scoring summaries at --v=4 ([#99411](https://github.com/kubernetes/kubernetes/pull/99411), [@damemi](https://github.com/damemi)) [SIG Scheduling] -- Kubeadm now includes CoreDNS v1.8.0. ([#96429](https://github.com/kubernetes/kubernetes/pull/96429), [@rajansandeep](https://github.com/rajansandeep)) [SIG Cluster Lifecycle] -- Kubeadm: IPv6DualStack feature gate graduates to Beta and enabled by default ([#99294](https://github.com/kubernetes/kubernetes/pull/99294), [@pacoxu](https://github.com/pacoxu)) -- Kubeadm: a warning to user as ipv6 site-local is deprecated ([#99574](https://github.com/kubernetes/kubernetes/pull/99574), [@pacoxu](https://github.com/pacoxu)) [SIG Cluster Lifecycle and Network] -- Kubeadm: add support for certificate chain validation. When using kubeadm in external CA mode, this allows an intermediate CA to be used to sign the certificates. The intermediate CA certificate must be appended to each signed certificate for this to work correctly. ([#97266](https://github.com/kubernetes/kubernetes/pull/97266), [@robbiemcmichael](https://github.com/robbiemcmichael)) [SIG Cluster Lifecycle] -- Kubeadm: amend the node kernel validation to treat CGROUP_PIDS, FAIR_GROUP_SCHED as required and CFS_BANDWIDTH, CGROUP_HUGETLB as optional ([#96378](https://github.com/kubernetes/kubernetes/pull/96378), [@neolit123](https://github.com/neolit123)) [SIG Cluster Lifecycle and Node] -- Kubeadm: apply the "node.kubernetes.io/exclude-from-external-load-balancers" label on control plane nodes during "init", "join" and "upgrade" to preserve backwards compatibility with the lagacy LB mode where nodes labeled as "master" where excluded. To opt-out you can remove the label from a node. See #97543 and the linked KEP for more details. ([#98269](https://github.com/kubernetes/kubernetes/pull/98269), [@neolit123](https://github.com/neolit123)) [SIG Cluster Lifecycle] -- Kubeadm: if the user has customized their image repository via the kubeadm configuration, pass the custom pause image repository and tag to the kubelet via --pod-infra-container-image not only for Docker but for all container runtimes. This flag tells the kubelet that it should not garbage collect the image. ([#99476](https://github.com/kubernetes/kubernetes/pull/99476), [@neolit123](https://github.com/neolit123)) [SIG Cluster Lifecycle] -- Kubeadm: perform pre-flight validation on host/node name upon `kubeadm init` and `kubeadm join`, showing warnings on non-compliant names ([#99194](https://github.com/kubernetes/kubernetes/pull/99194), [@pacoxu](https://github.com/pacoxu)) -- Kubectl version changed to write a warning message to stderr if the client and server version difference exceeds the supported version skew of +/-1 minor version. ([#98250](https://github.com/kubernetes/kubernetes/pull/98250), [@brianpursley](https://github.com/brianpursley)) [SIG CLI] -- Kubectl: Add `--use-protocol-buffers` flag to kubectl top pods and nodes. ([#96655](https://github.com/kubernetes/kubernetes/pull/96655), [@serathius](https://github.com/serathius)) -- Kubectl: `kubectl get` will omit managed fields by default now. Users could set `--show-managed-fields` to true to show managedFields when the output format is either `json` or `yaml`. ([#96878](https://github.com/kubernetes/kubernetes/pull/96878), [@knight42](https://github.com/knight42)) [SIG CLI and Testing] -- Kubectl: a Pod can be preselected as default container using `kubectl.kubernetes.io/default-container` annotation ([#99833](https://github.com/kubernetes/kubernetes/pull/99833), [@mengjiao-liu](https://github.com/mengjiao-liu)) -- Kubectl: add bash-completion for comma separated list on `kubectl get` ([#98301](https://github.com/kubernetes/kubernetes/pull/98301), [@phil9909](https://github.com/phil9909)) -- Kubernetes is now built using go1.15.8 ([#98834](https://github.com/kubernetes/kubernetes/pull/98834), [@cpanato](https://github.com/cpanato)) [SIG Cloud Provider, Instrumentation, Release and Testing] -- Kubernetes is now built with Golang 1.16 ([#98572](https://github.com/kubernetes/kubernetes/pull/98572), [@justaugustus](https://github.com/justaugustus)) [SIG API Machinery, Auth, CLI, Cloud Provider, Cluster Lifecycle, Instrumentation, Node, Release and Testing] -- Kubernetes is now built with Golang 1.16.1 ([#100106](https://github.com/kubernetes/kubernetes/pull/100106), [@justaugustus](https://github.com/justaugustus)) [SIG Cloud Provider, Instrumentation, Release and Testing] -- Metrics can now be disabled explicitly via a command line flag (i.e. '--disabled-metrics=metric1,metric2') ([#99217](https://github.com/kubernetes/kubernetes/pull/99217), [@logicalhan](https://github.com/logicalhan)) -- New admission controller `DenyServiceExternalIPs` is available. Clusters which do not *need* the Service `externalIPs` feature should enable this controller and be more secure. ([#97395](https://github.com/kubernetes/kubernetes/pull/97395), [@thockin](https://github.com/thockin)) -- Overall, enable the feature of `PreferNominatedNode` will improve the performance of scheduling where preemption might frequently happen, but in theory, enable the feature of `PreferNominatedNode`, the pod might not be scheduled to the best candidate node in the cluster. ([#93179](https://github.com/kubernetes/kubernetes/pull/93179), [@chendave](https://github.com/chendave)) [SIG Scheduling and Testing] -- Persistent Volumes formatted with the btrfs filesystem will now automatically resize when expanded. ([#99361](https://github.com/kubernetes/kubernetes/pull/99361), [@Novex](https://github.com/Novex)) [SIG Storage] -- Port the devicemanager to Windows node to allow device plugins like directx ([#93285](https://github.com/kubernetes/kubernetes/pull/93285), [@aarnaud](https://github.com/aarnaud)) [SIG Node, Testing and Windows] -- Removes cAdvisor JSON metrics (/stats/container, /stats//, /stats////) from the kubelet. ([#99236](https://github.com/kubernetes/kubernetes/pull/99236), [@pacoxu](https://github.com/pacoxu)) -- Rename metrics `etcd_object_counts` to `apiserver_storage_object_counts` and mark it as stable. The original `etcd_object_counts` metrics name is marked as "Deprecated" and will be removed in the future. ([#99785](https://github.com/kubernetes/kubernetes/pull/99785), [@erain](https://github.com/erain)) [SIG API Machinery, Instrumentation and Testing] -- Sysctls graduates to General Availability and thus unconditionally enabled. ([#99158](https://github.com/kubernetes/kubernetes/pull/99158), [@wgahnagl](https://github.com/wgahnagl)) -- The Kubernetes pause image manifest list now contains an image for Windows Server 20H2. ([#97322](https://github.com/kubernetes/kubernetes/pull/97322), [@claudiubelu](https://github.com/claudiubelu)) [SIG Windows] -- The NodeAffinity plugin implements the PreFilter extension, offering enhanced performance for Filter. ([#99213](https://github.com/kubernetes/kubernetes/pull/99213), [@AliceZhang2016](https://github.com/AliceZhang2016)) [SIG Scheduling] -- The `CronJobControllerV2` feature flag graduates to Beta and set to be enabled by default. ([#98878](https://github.com/kubernetes/kubernetes/pull/98878), [@soltysh](https://github.com/soltysh)) -- The `EndpointSlice` mirroring controller mirrors endpoints annotations and labels to the generated endpoint slices, it also ensures that updates on any of these fields are mirrored. - The well-known annotation `endpoints.kubernetes.io/last-change-trigger-time` is skipped and not mirrored. ([#98116](https://github.com/kubernetes/kubernetes/pull/98116), [@aojea](https://github.com/aojea)) -- The `RunAsGroup` feature has been promoted to GA in this release. ([#94641](https://github.com/kubernetes/kubernetes/pull/94641), [@krmayankk](https://github.com/krmayankk)) [SIG Auth and Node] -- The `ServiceAccountIssuerDiscovery` feature has graduated to GA, and is unconditionally enabled. The `ServiceAccountIssuerDiscovery` feature-gate will be removed in 1.22. ([#98553](https://github.com/kubernetes/kubernetes/pull/98553), [@mtaufen](https://github.com/mtaufen)) [SIG API Machinery, Auth and Testing] -- The `TTLAfterFinished` feature flag is now beta and enabled by default ([#98678](https://github.com/kubernetes/kubernetes/pull/98678), [@ahg-g](https://github.com/ahg-g)) -- The apimachinery util/net function used to detect the bind address `ResolveBindAddress()` takes into consideration global IP addresses on loopback interfaces when 1) the host has default routes, or 2) there are no global IPs on those interfaces in order to support more complex network scenarios like BGP Unnumbered RFC 5549 ([#95790](https://github.com/kubernetes/kubernetes/pull/95790), [@aojea](https://github.com/aojea)) [SIG Network] -- The feature gate `RootCAConfigMap` graduated to GA in v1.21 and therefore will be unconditionally enabled. This flag will be removed in v1.22 release. ([#98033](https://github.com/kubernetes/kubernetes/pull/98033), [@zshihang](https://github.com/zshihang)) -- The pause image upgraded to `v3.4.1` in kubelet and kubeadm for both Linux and Windows. ([#98205](https://github.com/kubernetes/kubernetes/pull/98205), [@pacoxu](https://github.com/pacoxu)) -- Update pause container to run as pseudo user and group `65535:65535`. This implies the release of version 3.5 of the container images. ([#97963](https://github.com/kubernetes/kubernetes/pull/97963), [@saschagrunert](https://github.com/saschagrunert)) [SIG CLI, Cloud Provider, Cluster Lifecycle, Node, Release, Security and Testing] -- Update the latest validated version of Docker to 20.10 ([#98977](https://github.com/kubernetes/kubernetes/pull/98977), [@neolit123](https://github.com/neolit123)) [SIG CLI, Cluster Lifecycle and Node] -- Upgrade node local dns to 1.17.0 for better IPv6 support ([#99749](https://github.com/kubernetes/kubernetes/pull/99749), [@pacoxu](https://github.com/pacoxu)) [SIG Cloud Provider and Network] -- Upgrades `IPv6Dualstack` to `Beta` and turns it on by default. New clusters or existing clusters are not be affected until an actor starts adding secondary Pods and service CIDRS CLI flags as described here: [IPv4/IPv6 Dual-stack](https://github.com/kubernetes/enhancements/tree/master/keps/sig-network/563-dual-stack) ([#98969](https://github.com/kubernetes/kubernetes/pull/98969), [@khenidak](https://github.com/khenidak)) -- Users might specify the `kubectl.kubernetes.io/default-container` annotation in a Pod to preselect container for kubectl commands. ([#99581](https://github.com/kubernetes/kubernetes/pull/99581), [@mengjiao-liu](https://github.com/mengjiao-liu)) [SIG CLI] -- When downscaling ReplicaSets, ready and creation timestamps are compared in a logarithmic scale. ([#99212](https://github.com/kubernetes/kubernetes/pull/99212), [@damemi](https://github.com/damemi)) [SIG Apps and Testing] -- When the kubelet is watching a ConfigMap or Secret purely in the context of setting environment variables - for containers, only hold that watch for a defined duration before cancelling it. This change reduces the CPU - and memory usage of the kube-apiserver in large clusters. ([#99393](https://github.com/kubernetes/kubernetes/pull/99393), [@chenyw1990](https://github.com/chenyw1990)) [SIG API Machinery, Node and Testing] -- WindowsEndpointSliceProxying feature gate has graduated to beta and is enabled by default. This means kube-proxy will read from EndpointSlices instead of Endpoints on Windows by default. ([#99794](https://github.com/kubernetes/kubernetes/pull/99794), [@robscott](https://github.com/robscott)) [SIG Network] -- `kubectl wait` ensures that observedGeneration >= generation to prevent stale state reporting. An example scenario can be found on CRD updates. ([#97408](https://github.com/kubernetes/kubernetes/pull/97408), [@KnicKnic](https://github.com/KnicKnic)) - -### 문서 - -- Azure file migration graduates to beta, with CSIMigrationAzureFile flag off by default - as it requires installation of AzureFile CSI Driver. Users should enable CSIMigration and - CSIMigrationAzureFile features and install the [AzureFile CSI Driver](https://github.com/kubernetes-sigs/azurefile-csi-driver) - to avoid disruption to existing Pod and PVC objects at that time. Azure File CSI driver does not support using same persistent - volume with different fsgroups. When CSI migration is enabled for azurefile driver, such case is not supported. - (there is a case we support where volume is mounted with 0777 and then it readable/writable by everyone) ([#96293](https://github.com/kubernetes/kubernetes/pull/96293), [@andyzhangx](https://github.com/andyzhangx)) -- Official support to build kubernetes with docker-machine / remote docker is removed. This change does not affect building kubernetes with docker locally. ([#97935](https://github.com/kubernetes/kubernetes/pull/97935), [@adeniyistephen](https://github.com/adeniyistephen)) [SIG Release and Testing] -- Set kubelet option `--volume-stats-agg-period` to negative value to disable volume calculations. ([#96675](https://github.com/kubernetes/kubernetes/pull/96675), [@pacoxu](https://github.com/pacoxu)) [SIG Node] - -### 실패 테스트 - -- Escape the special characters like `[`, `]` and ` ` that exist in vsphere windows path ([#98830](https://github.com/kubernetes/kubernetes/pull/98830), [@liyanhui1228](https://github.com/liyanhui1228)) [SIG Storage and Windows] -- Kube-proxy: fix a bug on UDP `NodePort` Services where stale connection tracking entries may blackhole the traffic directed to the `NodePort` ([#98305](https://github.com/kubernetes/kubernetes/pull/98305), [@aojea](https://github.com/aojea)) -- Kubelet: fixes a bug in the HostPort dockershim implementation that caused the conformance test "HostPort validates that there is no conflict between pods with same hostPort but different hostIP and protocol" to fail. ([#98755](https://github.com/kubernetes/kubernetes/pull/98755), [@aojea](https://github.com/aojea)) [SIG Cloud Provider, Network and Node] - -### 버그 또는 회귀(regression) - -- AcceleratorStats will be available in the Summary API of kubelet when cri_stats_provider is used. ([#96873](https://github.com/kubernetes/kubernetes/pull/96873), [@ruiwen-zhao](https://github.com/ruiwen-zhao)) [SIG Node] -- All data is no longer automatically deleted when a failure is detected during creation of the volume data file on a CSI volume. Now only the data file and volume path is removed. ([#96021](https://github.com/kubernetes/kubernetes/pull/96021), [@huffmanca](https://github.com/huffmanca)) -- Clean ReplicaSet by revision instead of creation timestamp in deployment controller ([#97407](https://github.com/kubernetes/kubernetes/pull/97407), [@waynepeking348](https://github.com/waynepeking348)) [SIG Apps] -- Cleanup subnet in frontend IP configs to prevent huge subnet request bodies in some scenarios. ([#98133](https://github.com/kubernetes/kubernetes/pull/98133), [@nilo19](https://github.com/nilo19)) [SIG Cloud Provider] -- Client-go exec credential plugins will pass stdin only when interactive terminal is detected on stdin. This fixes a bug where previously it was checking if **stdout** is an interactive terminal. ([#99654](https://github.com/kubernetes/kubernetes/pull/99654), [@ankeesler](https://github.com/ankeesler)) -- Cloud-controller-manager: routes controller should not depend on --allocate-node-cidrs ([#97029](https://github.com/kubernetes/kubernetes/pull/97029), [@andrewsykim](https://github.com/andrewsykim)) [SIG Cloud Provider and Testing] -- Cluster Autoscaler version bump to v1.20.0 ([#97011](https://github.com/kubernetes/kubernetes/pull/97011), [@towca](https://github.com/towca)) -- Creating a PVC with DataSource should fail for non-CSI plugins. ([#97086](https://github.com/kubernetes/kubernetes/pull/97086), [@xing-yang](https://github.com/xing-yang)) [SIG Apps and Storage] -- EndpointSlice controller is now less likely to emit FailedToUpdateEndpointSlices events. ([#99345](https://github.com/kubernetes/kubernetes/pull/99345), [@robscott](https://github.com/robscott)) [SIG Apps and Network] -- EndpointSlice controllers are less likely to create duplicate EndpointSlices. ([#100103](https://github.com/kubernetes/kubernetes/pull/100103), [@robscott](https://github.com/robscott)) [SIG Apps and Network] -- EndpointSliceMirroring controller is now less likely to emit FailedToUpdateEndpointSlices events. ([#99756](https://github.com/kubernetes/kubernetes/pull/99756), [@robscott](https://github.com/robscott)) [SIG Apps and Network] -- Ensure all vSphere nodes are are tracked by volume attach-detach controller ([#96689](https://github.com/kubernetes/kubernetes/pull/96689), [@gnufied](https://github.com/gnufied)) -- Ensure empty string annotations are copied over in rollbacks. ([#94858](https://github.com/kubernetes/kubernetes/pull/94858), [@waynepeking348](https://github.com/waynepeking348)) -- Ensure only one LoadBalancer rule is created when HA mode is enabled ([#99825](https://github.com/kubernetes/kubernetes/pull/99825), [@feiskyer](https://github.com/feiskyer)) [SIG Cloud Provider] -- Ensure that client-go's EventBroadcaster is safe (non-racy) during shutdown. ([#95664](https://github.com/kubernetes/kubernetes/pull/95664), [@DirectXMan12](https://github.com/DirectXMan12)) [SIG API Machinery] -- Explicitly pass `KUBE_BUILD_CONFORMANCE=y` in `package-tarballs` to reenable building the conformance tarballs. ([#100571](https://github.com/kubernetes/kubernetes/pull/100571), [@puerco](https://github.com/puerco)) -- Fix Azure file migration e2e test failure when CSIMigration is turned on. ([#97877](https://github.com/kubernetes/kubernetes/pull/97877), [@andyzhangx](https://github.com/andyzhangx)) -- Fix CSI-migrated inline EBS volumes failing to mount if their volumeID is prefixed by aws:// ([#96821](https://github.com/kubernetes/kubernetes/pull/96821), [@wongma7](https://github.com/wongma7)) [SIG Storage] -- Fix CVE-2020-8555 for Gluster client connections. ([#97922](https://github.com/kubernetes/kubernetes/pull/97922), [@liggitt](https://github.com/liggitt)) [SIG Storage] -- Fix NPE in ephemeral storage eviction ([#98261](https://github.com/kubernetes/kubernetes/pull/98261), [@wzshiming](https://github.com/wzshiming)) [SIG Node] -- Fix PermissionDenied issue on SMB mount for Windows ([#99550](https://github.com/kubernetes/kubernetes/pull/99550), [@andyzhangx](https://github.com/andyzhangx)) -- Fix bug that would let the Horizontal Pod Autoscaler scale down despite at least one metric being unavailable/invalid ([#99514](https://github.com/kubernetes/kubernetes/pull/99514), [@mikkeloscar](https://github.com/mikkeloscar)) [SIG Apps and Autoscaling] -- Fix cgroup handling for systemd with cgroup v2 ([#98365](https://github.com/kubernetes/kubernetes/pull/98365), [@odinuge](https://github.com/odinuge)) [SIG Node] -- Fix counting error in service/nodeport/loadbalancer quota check ([#97451](https://github.com/kubernetes/kubernetes/pull/97451), [@pacoxu](https://github.com/pacoxu)) [SIG API Machinery, Network and Testing] -- Fix errors when accessing Windows container stats for Dockershim ([#98510](https://github.com/kubernetes/kubernetes/pull/98510), [@jsturtevant](https://github.com/jsturtevant)) [SIG Node and Windows] -- Fix kube-proxy container image architecture for non amd64 images. ([#98526](https://github.com/kubernetes/kubernetes/pull/98526), [@saschagrunert](https://github.com/saschagrunert)) -- Fix missing cadvisor machine metrics. ([#97006](https://github.com/kubernetes/kubernetes/pull/97006), [@lingsamuel](https://github.com/lingsamuel)) [SIG Node] -- Fix nil VMSS name when setting service to auto mode ([#97366](https://github.com/kubernetes/kubernetes/pull/97366), [@nilo19](https://github.com/nilo19)) [SIG Cloud Provider] -- Fix privileged config of Pod Sandbox which was previously ignored. ([#96877](https://github.com/kubernetes/kubernetes/pull/96877), [@xeniumlee](https://github.com/xeniumlee)) -- Fix the panic when kubelet registers if a node object already exists with no Status.Capacity or Status.Allocatable ([#95269](https://github.com/kubernetes/kubernetes/pull/95269), [@SataQiu](https://github.com/SataQiu)) [SIG Node] -- Fix the regression with the slow pods termination. Before this fix pods may take an additional time to terminate - up to one minute. Reversing the change that ensured that CNI resources cleaned up when the pod is removed on API server. ([#97980](https://github.com/kubernetes/kubernetes/pull/97980), [@SergeyKanzhelev](https://github.com/SergeyKanzhelev)) [SIG Node] -- Fix to recover CSI volumes from certain dangling attachments ([#96617](https://github.com/kubernetes/kubernetes/pull/96617), [@yuga711](https://github.com/yuga711)) [SIG Apps and Storage] -- Fix: azure file latency issue for metadata-heavy workloads ([#97082](https://github.com/kubernetes/kubernetes/pull/97082), [@andyzhangx](https://github.com/andyzhangx)) [SIG Cloud Provider and Storage] -- Fixed Cinder volume IDs on OpenStack Train ([#96673](https://github.com/kubernetes/kubernetes/pull/96673), [@jsafrane](https://github.com/jsafrane)) [SIG Cloud Provider] -- Fixed FibreChannel volume plugin corrupting filesystems on detach of multipath volumes. ([#97013](https://github.com/kubernetes/kubernetes/pull/97013), [@jsafrane](https://github.com/jsafrane)) [SIG Storage] -- Fixed a bug in kubelet that will saturate CPU utilization after containerd got restarted. ([#97174](https://github.com/kubernetes/kubernetes/pull/97174), [@hanlins](https://github.com/hanlins)) [SIG Node] -- Fixed a bug that causes smaller number of conntrack-max being used under CPU static policy. (#99225, @xh4n3) ([#99613](https://github.com/kubernetes/kubernetes/pull/99613), [@xh4n3](https://github.com/xh4n3)) [SIG Network] -- Fixed a bug that on k8s nodes, when the policy of INPUT chain in filter table is not ACCEPT, healthcheck nodeport would not work. - Added iptables rules to allow healthcheck nodeport traffic. ([#97824](https://github.com/kubernetes/kubernetes/pull/97824), [@hanlins](https://github.com/hanlins)) [SIG Network] -- Fixed a bug that the kubelet cannot start on BtrfS. ([#98042](https://github.com/kubernetes/kubernetes/pull/98042), [@gjkim42](https://github.com/gjkim42)) [SIG Node] -- Fixed a race condition on API server startup ensuring previously created webhook configurations are effective before the first write request is admitted. ([#95783](https://github.com/kubernetes/kubernetes/pull/95783), [@roycaihw](https://github.com/roycaihw)) [SIG API Machinery] -- Fixed an issue with garbage collection failing to clean up namespaced children of an object also referenced incorrectly by cluster-scoped children ([#98068](https://github.com/kubernetes/kubernetes/pull/98068), [@liggitt](https://github.com/liggitt)) [SIG API Machinery and Apps] -- Fixed authentication_duration_seconds metric scope. Previously, it included whole apiserver request duration which yields inaccurate results. ([#99944](https://github.com/kubernetes/kubernetes/pull/99944), [@marseel](https://github.com/marseel)) -- Fixed bug in CPUManager with race on container map access ([#97427](https://github.com/kubernetes/kubernetes/pull/97427), [@klueska](https://github.com/klueska)) [SIG Node] -- Fixed bug that caused cAdvisor to incorrectly detect single-socket multi-NUMA topology. ([#99315](https://github.com/kubernetes/kubernetes/pull/99315), [@iwankgb](https://github.com/iwankgb)) [SIG Node] -- Fixed cleanup of block devices when /var/lib/kubelet is a symlink. ([#96889](https://github.com/kubernetes/kubernetes/pull/96889), [@jsafrane](https://github.com/jsafrane)) [SIG Storage] -- Fixed no effect namespace when exposing deployment with --dry-run=client. ([#97492](https://github.com/kubernetes/kubernetes/pull/97492), [@masap](https://github.com/masap)) [SIG CLI] -- Fixed provisioning of Cinder volumes migrated to CSI when StorageClass with AllowedTopologies was used. ([#98311](https://github.com/kubernetes/kubernetes/pull/98311), [@jsafrane](https://github.com/jsafrane)) [SIG Storage] -- Fixes a bug of identifying the correct containerd process. ([#97888](https://github.com/kubernetes/kubernetes/pull/97888), [@pacoxu](https://github.com/pacoxu)) -- Fixes add-on manager leader election to use leases instead of endpoints, similar to what kube-controller-manager does in 1.20 ([#98968](https://github.com/kubernetes/kubernetes/pull/98968), [@liggitt](https://github.com/liggitt)) -- Fixes connection errors when using `--volume-host-cidr-denylist` or `--volume-host-allow-local-loopback` ([#98436](https://github.com/kubernetes/kubernetes/pull/98436), [@liggitt](https://github.com/liggitt)) [SIG Network and Storage] -- Fixes problem where invalid selector on `PodDisruptionBudget` leads to a nil pointer dereference that causes the Controller manager to crash loop. ([#98750](https://github.com/kubernetes/kubernetes/pull/98750), [@mortent](https://github.com/mortent)) -- Fixes spurious errors about IPv6 in `kube-proxy` logs on nodes with IPv6 disabled. ([#99127](https://github.com/kubernetes/kubernetes/pull/99127), [@danwinship](https://github.com/danwinship)) -- Fixing a bug where a failed node may not have the NoExecute taint set correctly ([#96876](https://github.com/kubernetes/kubernetes/pull/96876), [@howieyuen](https://github.com/howieyuen)) [SIG Apps and Node] -- GCE Internal LoadBalancer sync loop will now release the ILB IP address upon sync failure. An error in ILB forwarding rule creation will no longer leak IP addresses. ([#97740](https://github.com/kubernetes/kubernetes/pull/97740), [@prameshj](https://github.com/prameshj)) [SIG Cloud Provider and Network] -- Ignore update pod with no new images in alwaysPullImages admission controller ([#96668](https://github.com/kubernetes/kubernetes/pull/96668), [@pacoxu](https://github.com/pacoxu)) [SIG Apps, Auth and Node] -- Improve speed of vSphere PV provisioning and reduce number of API calls ([#100054](https://github.com/kubernetes/kubernetes/pull/100054), [@gnufied](https://github.com/gnufied)) [SIG Cloud Provider and Storage] -- KUBECTL_EXTERNAL_DIFF now accepts equal sign for additional parameters. ([#98158](https://github.com/kubernetes/kubernetes/pull/98158), [@dougsland](https://github.com/dougsland)) [SIG CLI] -- Kube-apiserver: an update of a pod with a generic ephemeral volume dropped that volume if the feature had been disabled since creating the pod with such a volume ([#99446](https://github.com/kubernetes/kubernetes/pull/99446), [@pohly](https://github.com/pohly)) [SIG Apps, Node and Storage] -- Kube-proxy: remove deprecated --cleanup-ipvs flag of kube-proxy, and make --cleanup flag always to flush IPVS ([#97336](https://github.com/kubernetes/kubernetes/pull/97336), [@maaoBit](https://github.com/maaoBit)) [SIG Network] -- Kubeadm installs etcd v3.4.13 when creating cluster v1.19 ([#97244](https://github.com/kubernetes/kubernetes/pull/97244), [@pacoxu](https://github.com/pacoxu)) -- Kubeadm: Fixes a kubeadm upgrade bug that could cause a custom CoreDNS configuration to be replaced with the default. ([#97016](https://github.com/kubernetes/kubernetes/pull/97016), [@rajansandeep](https://github.com/rajansandeep)) [SIG Cluster Lifecycle] -- Kubeadm: Some text in the `kubeadm upgrade plan` output has changed. If you have scripts or other automation that parses this output, please review these changes and update your scripts to account for the new output. ([#98728](https://github.com/kubernetes/kubernetes/pull/98728), [@stmcginnis](https://github.com/stmcginnis)) [SIG Cluster Lifecycle] -- Kubeadm: fix a bug in the host memory detection code on 32bit Linux platforms ([#97403](https://github.com/kubernetes/kubernetes/pull/97403), [@abelbarrera15](https://github.com/abelbarrera15)) [SIG Cluster Lifecycle] -- Kubeadm: fix a bug where "kubeadm join" would not properly handle missing names for existing etcd members. ([#97372](https://github.com/kubernetes/kubernetes/pull/97372), [@ihgann](https://github.com/ihgann)) [SIG Cluster Lifecycle] -- Kubeadm: fix a bug where "kubeadm upgrade" commands can fail if CoreDNS v1.8.0 is installed. ([#97919](https://github.com/kubernetes/kubernetes/pull/97919), [@neolit123](https://github.com/neolit123)) [SIG Cluster Lifecycle] -- Kubeadm: fix a bug where external credentials in an existing admin.conf prevented the CA certificate to be written in the cluster-info ConfigMap. ([#98882](https://github.com/kubernetes/kubernetes/pull/98882), [@kvaps](https://github.com/kvaps)) [SIG Cluster Lifecycle] -- Kubeadm: get k8s CI version markers from k8s infra bucket ([#98836](https://github.com/kubernetes/kubernetes/pull/98836), [@hasheddan](https://github.com/hasheddan)) [SIG Cluster Lifecycle and Release] -- Kubeadm: skip validating pod subnet against node-cidr-mask when allocate-node-cidrs is set to be false ([#98984](https://github.com/kubernetes/kubernetes/pull/98984), [@SataQiu](https://github.com/SataQiu)) [SIG Cluster Lifecycle] -- Kubectl logs: `--ignore-errors` is now honored by all containers, maintaining consistency with parallelConsumeRequest behavior. ([#97686](https://github.com/kubernetes/kubernetes/pull/97686), [@wzshiming](https://github.com/wzshiming)) -- Kubectl-convert: Fix `no kind "Ingress" is registered for version` error ([#97754](https://github.com/kubernetes/kubernetes/pull/97754), [@wzshiming](https://github.com/wzshiming)) -- Kubectl: Fixed panic when describing an ingress backend without an API Group ([#100505](https://github.com/kubernetes/kubernetes/pull/100505), [@lauchokyip](https://github.com/lauchokyip)) [SIG CLI] -- Kubelet now cleans up orphaned volume directories automatically ([#95301](https://github.com/kubernetes/kubernetes/pull/95301), [@lorenz](https://github.com/lorenz)) [SIG Node and Storage] -- Kubelet.exe on Windows now checks that the process running as administrator and the executing user account is listed in the built-in administrators group. This is the equivalent to checking the process is running as uid 0. ([#96616](https://github.com/kubernetes/kubernetes/pull/96616), [@perithompson](https://github.com/perithompson)) [SIG Node and Windows] -- Kubelet: Fix kubelet from panic after getting the wrong signal ([#98200](https://github.com/kubernetes/kubernetes/pull/98200), [@wzshiming](https://github.com/wzshiming)) [SIG Node] -- Kubelet: Fix repeatedly acquiring the inhibit lock ([#98088](https://github.com/kubernetes/kubernetes/pull/98088), [@wzshiming](https://github.com/wzshiming)) [SIG Node] -- Kubelet: Fixed the bug of getting the number of cpu when the number of cpu logical processors is more than 64 in windows ([#97378](https://github.com/kubernetes/kubernetes/pull/97378), [@hwdef](https://github.com/hwdef)) [SIG Node and Windows] -- Limits lease to have 1000 maximum attached objects. ([#98257](https://github.com/kubernetes/kubernetes/pull/98257), [@lingsamuel](https://github.com/lingsamuel)) -- Mitigate CVE-2020-8555 for kube-up using GCE by preventing local loopback folume hosts. ([#97934](https://github.com/kubernetes/kubernetes/pull/97934), [@mattcary](https://github.com/mattcary)) [SIG Cloud Provider and Storage] -- On single-stack configured (IPv4 or IPv6, but not both) clusters, Services which are both headless (no clusterIP) and selectorless (empty or undefined selector) will report `ipFamilyPolicy RequireDualStack` and will have entries in `ipFamilies[]` for both IPv4 and IPv6. This is a change from alpha, but does not have any impact on the manually-specified Endpoints and EndpointSlices for the Service. ([#99555](https://github.com/kubernetes/kubernetes/pull/99555), [@thockin](https://github.com/thockin)) [SIG Apps and Network] -- Performance regression #97685 has been fixed. ([#97860](https://github.com/kubernetes/kubernetes/pull/97860), [@MikeSpreitzer](https://github.com/MikeSpreitzer)) [SIG API Machinery] -- Pod Log stats for windows now reports metrics ([#99221](https://github.com/kubernetes/kubernetes/pull/99221), [@jsturtevant](https://github.com/jsturtevant)) [SIG Node, Storage, Testing and Windows] -- Pod status updates faster when reacting on probe results. The first readiness probe will be called faster when startup probes succeeded, which will make Pod status as ready faster. ([#98376](https://github.com/kubernetes/kubernetes/pull/98376), [@matthyx](https://github.com/matthyx)) -- Readjust `kubelet_containers_per_pod_count` buckets to only show metrics greater than 1. ([#98169](https://github.com/kubernetes/kubernetes/pull/98169), [@wawa0210](https://github.com/wawa0210)) -- Remove CSI topology from migrated in-tree gcepd volume. ([#97823](https://github.com/kubernetes/kubernetes/pull/97823), [@Jiawei0227](https://github.com/Jiawei0227)) [SIG Cloud Provider and Storage] -- Requests with invalid timeout parameters in the request URL now appear in the audit log correctly. ([#96901](https://github.com/kubernetes/kubernetes/pull/96901), [@tkashem](https://github.com/tkashem)) [SIG API Machinery and Testing] -- Resolve a "concurrent map read and map write" crashing error in the kubelet ([#95111](https://github.com/kubernetes/kubernetes/pull/95111), [@choury](https://github.com/choury)) [SIG Node] -- Resolves spurious `Failed to list *v1.Secret` or `Failed to list *v1.ConfigMap` messages in kubelet logs. ([#99538](https://github.com/kubernetes/kubernetes/pull/99538), [@liggitt](https://github.com/liggitt)) [SIG Auth and Node] -- ResourceQuota of an entity now inclusively calculate Pod overhead ([#99600](https://github.com/kubernetes/kubernetes/pull/99600), [@gjkim42](https://github.com/gjkim42)) -- Return zero time (midnight on Jan. 1, 1970) instead of negative number when reporting startedAt and finishedAt of the not started or a running Pod when using `dockershim` as a runtime. ([#99585](https://github.com/kubernetes/kubernetes/pull/99585), [@Iceber](https://github.com/Iceber)) -- Reverts breaking change to inline AzureFile volumes; referenced secrets are now searched for in the same namespace as the pod as in previous releases. ([#100563](https://github.com/kubernetes/kubernetes/pull/100563), [@msau42](https://github.com/msau42)) -- Scores from InterPodAffinity have stronger differentiation. ([#98096](https://github.com/kubernetes/kubernetes/pull/98096), [@leileiwan](https://github.com/leileiwan)) [SIG Scheduling] -- Specifying the KUBE_TEST_REPO environment variable when e2e tests are executed will instruct the test infrastructure to load that image from a location within the specified repo, using a predefined pattern. ([#93510](https://github.com/kubernetes/kubernetes/pull/93510), [@smarterclayton](https://github.com/smarterclayton)) [SIG Testing] -- Static pods will be deleted gracefully. ([#98103](https://github.com/kubernetes/kubernetes/pull/98103), [@gjkim42](https://github.com/gjkim42)) [SIG Node] -- Sync node status during kubelet node shutdown. - Adds an pod admission handler that rejects new pods when the node is in progress of shutting down. ([#98005](https://github.com/kubernetes/kubernetes/pull/98005), [@wzshiming](https://github.com/wzshiming)) [SIG Node] -- The calculation of pod UIDs for static pods has changed to ensure each static pod gets a unique value - this will cause all static pod containers to be recreated/restarted if an in-place kubelet upgrade from 1.20 to 1.21 is performed. Note that draining pods before upgrading the kubelet across minor versions is the supported upgrade path. ([#87461](https://github.com/kubernetes/kubernetes/pull/87461), [@bboreham](https://github.com/bboreham)) [SIG Node] -- The maximum number of ports allowed in EndpointSlices has been increased from 100 to 20,000 ([#99795](https://github.com/kubernetes/kubernetes/pull/99795), [@robscott](https://github.com/robscott)) [SIG Network] -- Truncates a message if it hits the `NoteLengthLimit` when the scheduler records an event for the pod that indicates the pod has failed to schedule. ([#98715](https://github.com/kubernetes/kubernetes/pull/98715), [@carlory](https://github.com/carlory)) -- Updated k8s.gcr.io/ingress-gce-404-server-with-metrics-amd64 to a version that serves /metrics endpoint on a non-default port. ([#97621](https://github.com/kubernetes/kubernetes/pull/97621), [@vbannai](https://github.com/vbannai)) [SIG Cloud Provider] -- Updates the commands ` - - kubectl kustomize {arg} - - kubectl apply -k {arg} - `to use same code as kustomize CLI [v4.0.5](https://github.com/kubernetes-sigs/kustomize/releases/tag/kustomize%2Fv4.0.5) ([#98946](https://github.com/kubernetes/kubernetes/pull/98946), [@monopole](https://github.com/monopole)) -- Use force unmount for NFS volumes if regular mount fails after 1 minute timeout ([#96844](https://github.com/kubernetes/kubernetes/pull/96844), [@gnufied](https://github.com/gnufied)) [SIG Storage] -- Use network.Interface.VirtualMachine.ID to get the binded VM - Skip standalone VM when reconciling LoadBalancer ([#97635](https://github.com/kubernetes/kubernetes/pull/97635), [@nilo19](https://github.com/nilo19)) [SIG Cloud Provider] -- Using exec auth plugins with kubectl no longer results in warnings about constructing many client instances from the same exec auth config. ([#97857](https://github.com/kubernetes/kubernetes/pull/97857), [@liggitt](https://github.com/liggitt)) [SIG API Machinery and Auth] -- When a CNI plugin returns dual-stack pod IPs, kubelet will now try to respect the - "primary IP family" of the cluster by picking a primary pod IP of the same family - as the (primary) node IP, rather than assuming that the CNI plugin returned the IPs - in the order the administrator wanted (since some CNI plugins don't allow - configuring this). ([#97979](https://github.com/kubernetes/kubernetes/pull/97979), [@danwinship](https://github.com/danwinship)) [SIG Network and Node] -- When dynamically provisioning Azure File volumes for a premium account, the requested size will be set to 100GB if the request is initially lower than this value to accommodate Azure File requirements. ([#99122](https://github.com/kubernetes/kubernetes/pull/99122), [@huffmanca](https://github.com/huffmanca)) [SIG Cloud Provider and Storage] -- When using `Containerd` on Windows, the `C:\Windows\System32\drivers\etc\hosts` file will now be managed by kubelet. ([#83730](https://github.com/kubernetes/kubernetes/pull/83730), [@claudiubelu](https://github.com/claudiubelu)) -- `VolumeBindingArgs` now allow `BindTimeoutSeconds` to be set as zero, while the value zero indicates no waiting for the checking of volume binding operation. ([#99835](https://github.com/kubernetes/kubernetes/pull/99835), [@chendave](https://github.com/chendave)) [SIG Scheduling and Storage] -- `kubectl exec` and `kubectl attach` now honor the `--quiet` flag which suppresses output from the local binary that could be confused by a script with the remote command output (all non-failure output is hidden). In addition, print inline with exec and attach the list of alternate containers when we default to the first spec.container. ([#99004](https://github.com/kubernetes/kubernetes/pull/99004), [@smarterclayton](https://github.com/smarterclayton)) [SIG CLI] - -### 기타 (정리 또는 플레이크(flake)) - -- APIs for kubelet annotations and labels from `k8s.io/kubernetes/pkg/kubelet/apis` are now moved under `k8s.io/kubelet/pkg/apis/` ([#98931](https://github.com/kubernetes/kubernetes/pull/98931), [@michaelbeaumont](https://github.com/michaelbeaumont)) -- Apiserver_request_duration_seconds is promoted to stable status. ([#99925](https://github.com/kubernetes/kubernetes/pull/99925), [@logicalhan](https://github.com/logicalhan)) [SIG API Machinery, Instrumentation and Testing] -- Bump github.com/Azure/go-autorest/autorest to v0.11.12 ([#97033](https://github.com/kubernetes/kubernetes/pull/97033), [@patrickshan](https://github.com/patrickshan)) [SIG API Machinery, CLI, Cloud Provider and Cluster Lifecycle] -- Clients required to use go1.15.8+ or go1.16+ if kube-apiserver has the goaway feature enabled to avoid unexpected data race condition. ([#98809](https://github.com/kubernetes/kubernetes/pull/98809), [@answer1991](https://github.com/answer1991)) -- Delete deprecated `service.beta.kubernetes.io/azure-load-balancer-mixed-protocols` mixed procotol annotation in favor of the MixedProtocolLBService feature ([#97096](https://github.com/kubernetes/kubernetes/pull/97096), [@nilo19](https://github.com/nilo19)) [SIG Cloud Provider] -- EndpointSlice generation is now incremented when labels change. ([#99750](https://github.com/kubernetes/kubernetes/pull/99750), [@robscott](https://github.com/robscott)) [SIG Network] -- Featuregate AllowInsecureBackendProxy graduates to GA and unconditionally enabled. ([#99658](https://github.com/kubernetes/kubernetes/pull/99658), [@deads2k](https://github.com/deads2k)) -- Increase timeout for pod lifecycle test to reach pod status=ready ([#96691](https://github.com/kubernetes/kubernetes/pull/96691), [@hh](https://github.com/hh)) -- Increased `CSINodeIDMaxLength` from 128 bytes to 192 bytes. ([#98753](https://github.com/kubernetes/kubernetes/pull/98753), [@Jiawei0227](https://github.com/Jiawei0227)) -- Kube-apiserver: The OIDC authenticator no longer waits 10 seconds before attempting to fetch the metadata required to verify tokens. ([#97693](https://github.com/kubernetes/kubernetes/pull/97693), [@enj](https://github.com/enj)) [SIG API Machinery and Auth] -- Kube-proxy: Traffic from the cluster directed to ExternalIPs is always sent directly to the Service. ([#96296](https://github.com/kubernetes/kubernetes/pull/96296), [@aojea](https://github.com/aojea)) [SIG Network and Testing] -- Kubeadm: change the default image repository for CI images from 'gcr.io/kubernetes-ci-images' to 'gcr.io/k8s-staging-ci-images' ([#97087](https://github.com/kubernetes/kubernetes/pull/97087), [@SataQiu](https://github.com/SataQiu)) [SIG Cluster Lifecycle] -- Kubectl: The deprecated `kubectl alpha debug` command is removed. Use `kubectl debug` instead. ([#98111](https://github.com/kubernetes/kubernetes/pull/98111), [@pandaamanda](https://github.com/pandaamanda)) [SIG CLI] -- Kubelet command line flags related to dockershim are now showing deprecation message as they will be removed along with dockershim in future release. ([#98730](https://github.com/kubernetes/kubernetes/pull/98730), [@dims](https://github.com/dims)) -- Official support to build kubernetes with docker-machine / remote docker is removed. This change does not affect building kubernetes with docker locally. ([#97618](https://github.com/kubernetes/kubernetes/pull/97618), [@jherrera123](https://github.com/jherrera123)) [SIG Release and Testing] -- Process start time on Windows now uses current process information ([#97491](https://github.com/kubernetes/kubernetes/pull/97491), [@jsturtevant](https://github.com/jsturtevant)) [SIG API Machinery, CLI, Cloud Provider, Cluster Lifecycle, Instrumentation and Windows] -- Resolves flakes in the Ingress conformance tests due to conflicts with controllers updating the Ingress object ([#98430](https://github.com/kubernetes/kubernetes/pull/98430), [@liggitt](https://github.com/liggitt)) [SIG Network and Testing] -- The `AttachVolumeLimit` feature gate (GA since v1.17) has been removed and now unconditionally enabled. ([#96539](https://github.com/kubernetes/kubernetes/pull/96539), [@ialidzhikov](https://github.com/ialidzhikov)) -- The `CSINodeInfo` feature gate that is GA since v1.17 is unconditionally enabled, and can no longer be specified via the `--feature-gates` argument. ([#96561](https://github.com/kubernetes/kubernetes/pull/96561), [@ialidzhikov](https://github.com/ialidzhikov)) [SIG Apps, Auth, Scheduling, Storage and Testing] -- The `apiserver_request_total` metric is promoted to stable status and no longer has a content-type dimensions, so any alerts/charts which presume the existence of this will fail. This is however, unlikely to be the case since it was effectively an unbounded dimension in the first place. ([#99788](https://github.com/kubernetes/kubernetes/pull/99788), [@logicalhan](https://github.com/logicalhan)) -- The default delegating authorization options now allow unauthenticated access to healthz, readyz, and livez. A system:masters user connecting to an authz delegator will not perform an authz check. ([#98325](https://github.com/kubernetes/kubernetes/pull/98325), [@deads2k](https://github.com/deads2k)) [SIG API Machinery, Auth, Cloud Provider and Scheduling] -- The deprecated feature gates `CSIDriverRegistry`, `BlockVolume` and `CSIBlockVolume` are now unconditionally enabled and can no longer be specified in component invocations. ([#98021](https://github.com/kubernetes/kubernetes/pull/98021), [@gavinfish](https://github.com/gavinfish)) [SIG Storage] -- The deprecated feature gates `RotateKubeletClientCertificate`, `AttachVolumeLimit`, `VolumePVCDataSource` and `EvenPodsSpread` are now unconditionally enabled and can no longer be specified in component invocations. ([#97306](https://github.com/kubernetes/kubernetes/pull/97306), [@gavinfish](https://github.com/gavinfish)) [SIG Node, Scheduling and Storage] -- The e2e suite can be instructed not to wait for pods in kube-system to be ready or for all nodes to be ready by passing `--allowed-not-ready-nodes=-1` when invoking the e2e.test program. This allows callers to run subsets of the e2e suite in scenarios other than perfectly healthy clusters. ([#98781](https://github.com/kubernetes/kubernetes/pull/98781), [@smarterclayton](https://github.com/smarterclayton)) [SIG Testing] -- The feature gates `WindowsGMSA` and `WindowsRunAsUserName` that are GA since v1.18 are now removed. ([#96531](https://github.com/kubernetes/kubernetes/pull/96531), [@ialidzhikov](https://github.com/ialidzhikov)) [SIG Node and Windows] -- The new `-gce-zones` flag on the `e2e.test` binary instructs tests that check for information about how the cluster interacts with the cloud to limit their queries to the provided zone list. If not specified, the current behavior of asking the cloud provider for all available zones in multi zone clusters is preserved. ([#98787](https://github.com/kubernetes/kubernetes/pull/98787), [@smarterclayton](https://github.com/smarterclayton)) [SIG API Machinery, Cluster Lifecycle and Testing] -- Update cri-tools to [v1.20.0](https://github.com/kubernetes-sigs/cri-tools/releases/tag/v1.20.0) ([#97967](https://github.com/kubernetes/kubernetes/pull/97967), [@rajibmitra](https://github.com/rajibmitra)) [SIG Cloud Provider] -- Windows nodes on GCE will take longer to start due to dependencies installed at node creation time. ([#98284](https://github.com/kubernetes/kubernetes/pull/98284), [@pjh](https://github.com/pjh)) [SIG Cloud Provider] -- `apiserver_storage_objects` (a newer version of `etcd_object_counts`) is promoted and marked as stable. ([#100082](https://github.com/kubernetes/kubernetes/pull/100082), [@logicalhan](https://github.com/logicalhan)) - -### 분류되지 않음 - -- GCE L4 Loadbalancers now handle > 5 ports in service spec correctly. ([#99595](https://github.com/kubernetes/kubernetes/pull/99595), [@prameshj](https://github.com/prameshj)) [SIG Cloud Provider] -- The DownwardAPIHugePages feature is beta. Users may use the feature if all workers in their cluster are min 1.20 version. The feature will be enabled by default in all installations in 1.22. ([#99610](https://github.com/kubernetes/kubernetes/pull/99610), [@derekwaynecarr](https://github.com/derekwaynecarr)) [SIG Node] - -## 의존성 - -### 추가 -- github.com/go-errors/errors: [v1.0.1](https://github.com/go-errors/errors/tree/v1.0.1) -- github.com/gobuffalo/here: [v0.6.0](https://github.com/gobuffalo/here/tree/v0.6.0) -- github.com/google/shlex: [e7afc7f](https://github.com/google/shlex/tree/e7afc7f) -- github.com/markbates/pkger: [v0.17.1](https://github.com/markbates/pkger/tree/v0.17.1) -- github.com/moby/spdystream: [v0.2.0](https://github.com/moby/spdystream/tree/v0.2.0) -- github.com/monochromegane/go-gitignore: [205db1a](https://github.com/monochromegane/go-gitignore/tree/205db1a) -- github.com/niemeyer/pretty: [a10e7ca](https://github.com/niemeyer/pretty/tree/a10e7ca) -- github.com/xlab/treeprint: [a009c39](https://github.com/xlab/treeprint/tree/a009c39) -- go.starlark.net: 8dd3e2e -- golang.org/x/term: 6a3ed07 -- sigs.k8s.io/kustomize/api: v0.8.5 -- sigs.k8s.io/kustomize/cmd/config: v0.9.7 -- sigs.k8s.io/kustomize/kustomize/v4: v4.0.5 -- sigs.k8s.io/kustomize/kyaml: v0.10.15 - -### 변경 -- dmitri.shuralyov.com/gpu/mtl: 666a987 → 28db891 -- github.com/Azure/go-autorest/autorest: [v0.11.1 → v0.11.12](https://github.com/Azure/go-autorest/autorest/compare/v0.11.1...v0.11.12) -- github.com/NYTimes/gziphandler: [56545f4 → v1.1.1](https://github.com/NYTimes/gziphandler/compare/56545f4...v1.1.1) -- github.com/cilium/ebpf: [1c8d4c9 → v0.2.0](https://github.com/cilium/ebpf/compare/1c8d4c9...v0.2.0) -- github.com/container-storage-interface/spec: [v1.2.0 → v1.3.0](https://github.com/container-storage-interface/spec/compare/v1.2.0...v1.3.0) -- github.com/containerd/console: [v1.0.0 → v1.0.1](https://github.com/containerd/console/compare/v1.0.0...v1.0.1) -- github.com/containerd/containerd: [v1.4.1 → v1.4.4](https://github.com/containerd/containerd/compare/v1.4.1...v1.4.4) -- github.com/coredns/corefile-migration: [v1.0.10 → v1.0.11](https://github.com/coredns/corefile-migration/compare/v1.0.10...v1.0.11) -- github.com/creack/pty: [v1.1.7 → v1.1.11](https://github.com/creack/pty/compare/v1.1.7...v1.1.11) -- github.com/docker/docker: [bd33bbf → v20.10.2+incompatible](https://github.com/docker/docker/compare/bd33bbf...v20.10.2) -- github.com/go-logr/logr: [v0.2.0 → v0.4.0](https://github.com/go-logr/logr/compare/v0.2.0...v0.4.0) -- github.com/go-openapi/spec: [v0.19.3 → v0.19.5](https://github.com/go-openapi/spec/compare/v0.19.3...v0.19.5) -- github.com/go-openapi/strfmt: [v0.19.3 → v0.19.5](https://github.com/go-openapi/strfmt/compare/v0.19.3...v0.19.5) -- github.com/go-openapi/validate: [v0.19.5 → v0.19.8](https://github.com/go-openapi/validate/compare/v0.19.5...v0.19.8) -- github.com/gogo/protobuf: [v1.3.1 → v1.3.2](https://github.com/gogo/protobuf/compare/v1.3.1...v1.3.2) -- github.com/golang/mock: [v1.4.1 → v1.4.4](https://github.com/golang/mock/compare/v1.4.1...v1.4.4) -- github.com/google/cadvisor: [v0.38.5 → v0.39.0](https://github.com/google/cadvisor/compare/v0.38.5...v0.39.0) -- github.com/heketi/heketi: [c2e2a4a → v10.2.0+incompatible](https://github.com/heketi/heketi/compare/c2e2a4a...v10.2.0) -- github.com/kisielk/errcheck: [v1.2.0 → v1.5.0](https://github.com/kisielk/errcheck/compare/v1.2.0...v1.5.0) -- github.com/konsorten/go-windows-terminal-sequences: [v1.0.3 → v1.0.2](https://github.com/konsorten/go-windows-terminal-sequences/compare/v1.0.3...v1.0.2) -- github.com/kr/text: [v0.1.0 → v0.2.0](https://github.com/kr/text/compare/v0.1.0...v0.2.0) -- github.com/mattn/go-runewidth: [v0.0.2 → v0.0.7](https://github.com/mattn/go-runewidth/compare/v0.0.2...v0.0.7) -- github.com/miekg/dns: [v1.1.4 → v1.1.35](https://github.com/miekg/dns/compare/v1.1.4...v1.1.35) -- github.com/moby/sys/mountinfo: [v0.1.3 → v0.4.0](https://github.com/moby/sys/mountinfo/compare/v0.1.3...v0.4.0) -- github.com/moby/term: [672ec06 → df9cb8a](https://github.com/moby/term/compare/672ec06...df9cb8a) -- github.com/mrunalp/fileutils: [abd8a0e → v0.5.0](https://github.com/mrunalp/fileutils/compare/abd8a0e...v0.5.0) -- github.com/olekukonko/tablewriter: [a0225b3 → v0.0.4](https://github.com/olekukonko/tablewriter/compare/a0225b3...v0.0.4) -- github.com/opencontainers/runc: [v1.0.0-rc92 → v1.0.0-rc93](https://github.com/opencontainers/runc/compare/v1.0.0-rc92...v1.0.0-rc93) -- github.com/opencontainers/runtime-spec: [4d89ac9 → e6143ca](https://github.com/opencontainers/runtime-spec/compare/4d89ac9...e6143ca) -- github.com/opencontainers/selinux: [v1.6.0 → v1.8.0](https://github.com/opencontainers/selinux/compare/v1.6.0...v1.8.0) -- github.com/sergi/go-diff: [v1.0.0 → v1.1.0](https://github.com/sergi/go-diff/compare/v1.0.0...v1.1.0) -- github.com/sirupsen/logrus: [v1.6.0 → v1.7.0](https://github.com/sirupsen/logrus/compare/v1.6.0...v1.7.0) -- github.com/syndtr/gocapability: [d983527 → 42c35b4](https://github.com/syndtr/gocapability/compare/d983527...42c35b4) -- github.com/willf/bitset: [d5bec33 → v1.1.11](https://github.com/willf/bitset/compare/d5bec33...v1.1.11) -- github.com/yuin/goldmark: [v1.1.27 → v1.2.1](https://github.com/yuin/goldmark/compare/v1.1.27...v1.2.1) -- golang.org/x/crypto: 7f63de1 → 5ea612d -- golang.org/x/exp: 6cc2880 → 85be41e -- golang.org/x/mobile: d2bd2a2 → e6ae53a -- golang.org/x/mod: v0.3.0 → ce943fd -- golang.org/x/net: 69a7880 → 3d97a24 -- golang.org/x/sync: cd5d95a → 67f06af -- golang.org/x/sys: 5cba982 → a50acf3 -- golang.org/x/time: 3af7569 → f8bda1e -- golang.org/x/tools: c1934b7 → v0.1.0 -- gopkg.in/check.v1: 41f04d3 → 8fa4692 -- gopkg.in/yaml.v2: v2.2.8 → v2.4.0 -- gotest.tools/v3: v3.0.2 → v3.0.3 -- k8s.io/gengo: 83324d8 → b6c5ce2 -- k8s.io/klog/v2: v2.4.0 → v2.8.0 -- k8s.io/kube-openapi: d219536 → 591a79e -- k8s.io/system-validators: v1.2.0 → v1.4.0 -- sigs.k8s.io/apiserver-network-proxy/konnectivity-client: v0.0.14 → v0.0.15 -- sigs.k8s.io/structured-merge-diff/v4: v4.0.2 → v4.1.0 - -### 제거 -- github.com/codegangsta/negroni: [v1.0.0](https://github.com/codegangsta/negroni/tree/v1.0.0) -- github.com/docker/spdystream: [449fdfc](https://github.com/docker/spdystream/tree/449fdfc) -- github.com/golangplus/bytes: [45c989f](https://github.com/golangplus/bytes/tree/45c989f) -- github.com/golangplus/fmt: [2a5d6d7](https://github.com/golangplus/fmt/tree/2a5d6d7) -- github.com/gorilla/context: [v1.1.1](https://github.com/gorilla/context/tree/v1.1.1) -- github.com/kr/pty: [v1.1.5](https://github.com/kr/pty/tree/v1.1.5) -- rsc.io/quote/v3: v3.1.0 -- rsc.io/sampler: v1.3.0 -- sigs.k8s.io/kustomize: v2.0.3+incompatible - - - -# v1.21.0-rc.0 - - -## Downloads for v1.21.0-rc.0 - -### Source Code - -filename | sha512 hash --------- | ----------- -[kubernetes.tar.gz](https://dl.k8s.io/v1.21.0-rc.0/kubernetes.tar.gz) | ef53a41955d6f8a8d2a94636af98b55d633fb8a5081517559039e019b3dd65c9d10d4e7fa297ab88a7865d772f3eecf72e7b0eeba5e87accb4000c91da33e148 -[kubernetes-src.tar.gz](https://dl.k8s.io/v1.21.0-rc.0/kubernetes-src.tar.gz) | 9335a01b50d351776d3b8d00c07a5233844c51d307e361fa7e55a0620c1cb8b699e43eacf45ae9cafd8cbc44752e6987450c528a5bede8204706b7673000b5fc - -### Client binaries - -filename | sha512 hash --------- | ----------- -[kubernetes-client-darwin-amd64.tar.gz](https://dl.k8s.io/v1.21.0-rc.0/kubernetes-client-darwin-amd64.tar.gz) | 964135e43234cee275c452f5f06fb6d2bcd3cff3211a0d50fa35fff1cc4446bc5a0ac5125405dadcfb6596cb152afe29fabf7aad5b35b100e1288db890b70f8e -[kubernetes-client-darwin-arm64.tar.gz](https://dl.k8s.io/v1.21.0-rc.0/kubernetes-client-darwin-arm64.tar.gz) | 50d782abaa4ded5e706b3192d87effa953ceabbd7d91e3d48b0c1fa2206a1963a909c14b923560f5d09cac2c7392edc5f38a13fbf1e9a40bc94e3afe8de10622 -[kubernetes-client-linux-386.tar.gz](https://dl.k8s.io/v1.21.0-rc.0/kubernetes-client-linux-386.tar.gz) | 72af5562f24184a2d7c27f95fa260470da979fbdcacce39a372f8f3add2991d7af8bc78f4e1dbe7a0f97e3f559b149b72a51491d3b13008da81872ee50f02f37 -[kubernetes-client-linux-amd64.tar.gz](https://dl.k8s.io/v1.21.0-rc.0/kubernetes-client-linux-amd64.tar.gz) | 1eddb8f6b51e005bc6f7b519d036cbe3d2f6d97dbf7d212dd933fb56354c29f222d050519115a9bcf94555aef095db7cf763469e47bb4ae3c6c07f97edf437cb -[kubernetes-client-linux-arm.tar.gz](https://dl.k8s.io/v1.21.0-rc.0/kubernetes-client-linux-arm.tar.gz) | 670f8ca60ea3cf0bb3262a772715e0ea735fccda6a92f3186299361dc455b304ae177d4017e0b67bbfa4a95e36f4cc3f7eb335e2a5130c93ac3fba2aff4519bf -[kubernetes-client-linux-arm64.tar.gz](https://dl.k8s.io/v1.21.0-rc.0/kubernetes-client-linux-arm64.tar.gz) | a69a47907cff138ba393d8c87044fd95d97f3ca8f35d301b50742e2801ad7c229d99d6667971091f65825eb51854d585be0dd7421670110b1aa567e67e7ab4b3 -[kubernetes-client-linux-ppc64le.tar.gz](https://dl.k8s.io/v1.21.0-rc.0/kubernetes-client-linux-ppc64le.tar.gz) | b929feade94b71c81908abdcd4343b1e1e20098fd65e10d4d02585ad649d292d06f52c7ddc349efa188ce5b093e703c7aa9582c6ae5a69699adb87bbf5350243 -[kubernetes-client-linux-s390x.tar.gz](https://dl.k8s.io/v1.21.0-rc.0/kubernetes-client-linux-s390x.tar.gz) | 899d1470e412282cf289d8e24806d1a08c62ec0151f345ae3c9e497cc7bc0feab76498de4dd897d6adcdfa0c422e6b1a37e25d928669030f53457fd69d6e7df7 -[kubernetes-client-windows-386.tar.gz](https://dl.k8s.io/v1.21.0-rc.0/kubernetes-client-windows-386.tar.gz) | 9f0bc90a269eabd06fe4f637b5172a3a6a7d3de26de0d66504c2e1f2093083c584ea39031db6075a7da7a86b98c48bed25aa88d4ac09060b38692c6a5b637078 -[kubernetes-client-windows-amd64.tar.gz](https://dl.k8s.io/v1.21.0-rc.0/kubernetes-client-windows-amd64.tar.gz) | 05c8cc10188a1294b0d51d052942742a9b26411a08ec73494bf0e728a8a167e0a7863bdfc8864e76a371b584380098381805341e18b4b283b5d0cf298d5f7c7c - -### Server binaries - -filename | sha512 hash --------- | ----------- -[kubernetes-server-linux-amd64.tar.gz](https://dl.k8s.io/v1.21.0-rc.0/kubernetes-server-linux-amd64.tar.gz) | 355f278728ef7ac7eb2f5568c99c1429543c6302bbd0ed3bd0378c08116075e56ae850a49241313f078e2392702672ec6c9b70c8d97b4f2f5f4bee36828a63ba -[kubernetes-server-linux-arm.tar.gz](https://dl.k8s.io/v1.21.0-rc.0/kubernetes-server-linux-arm.tar.gz) | 9ac02c2825e2fd4e92f0c0f67180c67c24e32841ccbabc82284bf6293727ffecfae65e8a42b527c2a7ca482752384928eb65c2a1706144ae7819a6b3a1ab291c -[kubernetes-server-linux-arm64.tar.gz](https://dl.k8s.io/v1.21.0-rc.0/kubernetes-server-linux-arm64.tar.gz) | eb412453da03c82a9248412c8ccf4d4baa1fbfa81edd8d4f81d28969b40a3727e18934accc68f643d253446c58ffd2623292402495480b3d4b2a837b5318b957 -[kubernetes-server-linux-ppc64le.tar.gz](https://dl.k8s.io/v1.21.0-rc.0/kubernetes-server-linux-ppc64le.tar.gz) | 07da2812c35bbc427ee5b4a0b601c3ae271e0d50ab0dd4c5c25399f43506fa2a187642eb9d4d2085df7b90264d48ea2f31088af87d9efa7eb2e87f91e1fdbde4 -[kubernetes-server-linux-s390x.tar.gz](https://dl.k8s.io/v1.21.0-rc.0/kubernetes-server-linux-s390x.tar.gz) | 3b79442a3d6e389c4ff105922a8e49994c0b6c088d2c501bd8c78d9f9e814902f5bb72c8f9c89380b750fda9b3a336759b9b68f11d70bef4f0e984564a95c29e - -### Node binaries - -filename | sha512 hash --------- | ----------- -[kubernetes-node-linux-amd64.tar.gz](https://dl.k8s.io/v1.21.0-rc.0/kubernetes-node-linux-amd64.tar.gz) | f12edf1faf5f07de1ebc5a8626601c12927902e10aca3f11e398637382fdf55365dbd9a0ef38858553fb7569495ae2cf68f155dd2e49b85b27d76fb599bb92e4 -[kubernetes-node-linux-arm.tar.gz](https://dl.k8s.io/v1.21.0-rc.0/kubernetes-node-linux-arm.tar.gz) | 4fba8fc4e2102f07fb778aab597ec7231ea65c35e1aa618fe98b707b64a931237bd842c173e9120326e4d9deb983bb3917176762bba2212612bbc09d6e2105c4 -[kubernetes-node-linux-arm64.tar.gz](https://dl.k8s.io/v1.21.0-rc.0/kubernetes-node-linux-arm64.tar.gz) | a2e1be5459a8346839970faf4e7ebdb8ab9f3273e02babf1f3199b06bdb67434a2d18fcd1628cf1b989756e99d8dad6624a455b9db11d50f51f509f4df5c27da -[kubernetes-node-linux-ppc64le.tar.gz](https://dl.k8s.io/v1.21.0-rc.0/kubernetes-node-linux-ppc64le.tar.gz) | 16d2c1cc295474fc49fe9a827ddd73e81bdd6b76af7074987b90250023f99b6d70bf474e204c7d556802111984fcb3a330740b150bdc7970d0e3634eb94a1665 -[kubernetes-node-linux-s390x.tar.gz](https://dl.k8s.io/v1.21.0-rc.0/kubernetes-node-linux-s390x.tar.gz) | 9dc6faa6cd007b13dfce703f3e271f80adcc4e029c90a4a9b4f2f143b9756f2893f8af3d7c2cf813f2bd6731cffd87d15d4229456c1685939f65bf467820ec6e -[kubernetes-node-windows-amd64.tar.gz](https://dl.k8s.io/v1.21.0-rc.0/kubernetes-node-windows-amd64.tar.gz) | f8bac2974c9142bfb80cd5eadeda79f79f27b78899a4e6e71809b795c708824ba442be83fdbadb98e01c3823dd8350776358258a205e851ed045572923cacba7 - -## Changelog since v1.21.0-beta.1 - -## Urgent Upgrade Notes - -### (No, really, you MUST read this before you upgrade) - - - Migrated pkg/kubelet/cm/cpuset/cpuset.go to structured logging. Exit code changed from 255 to 1. ([#100007](https://github.com/kubernetes/kubernetes/pull/100007), [@utsavoza](https://github.com/utsavoza)) [SIG Instrumentation and Node] - -## Changes by Kind - -### API Change - -- Add Probe-level terminationGracePeriodSeconds field ([#99375](https://github.com/kubernetes/kubernetes/pull/99375), [@ehashman](https://github.com/ehashman)) [SIG API Machinery, Apps, Node and Testing] -- CSIServiceAccountToken is Beta now ([#99298](https://github.com/kubernetes/kubernetes/pull/99298), [@zshihang](https://github.com/zshihang)) [SIG Auth, Storage and Testing] -- Discovery.k8s.io/v1beta1 EndpointSlices are deprecated in favor of discovery.k8s.io/v1, and will no longer be served in Kubernetes v1.25. ([#100472](https://github.com/kubernetes/kubernetes/pull/100472), [@liggitt](https://github.com/liggitt)) [SIG Network] -- FieldManager no longer owns fields that get reset before the object is persisted (e.g. "status wiping"). ([#99661](https://github.com/kubernetes/kubernetes/pull/99661), [@kevindelgado](https://github.com/kevindelgado)) [SIG API Machinery, Auth and Testing] -- Generic ephemeral volumes are beta. ([#99643](https://github.com/kubernetes/kubernetes/pull/99643), [@pohly](https://github.com/pohly)) [SIG API Machinery, Apps, Auth, CLI, Node, Storage and Testing] -- Implement the GetAvailableResources in the podresources API. ([#95734](https://github.com/kubernetes/kubernetes/pull/95734), [@fromanirh](https://github.com/fromanirh)) [SIG Instrumentation, Node and Testing] -- The Endpoints controller will now set the `endpoints.kubernetes.io/over-capacity` annotation to "warning" when an Endpoints resource contains more than 1000 addresses. In a future release, the controller will truncate Endpoints that exceed this limit. The EndpointSlice API can be used to support significantly larger number of addresses. ([#99975](https://github.com/kubernetes/kubernetes/pull/99975), [@robscott](https://github.com/robscott)) [SIG Apps and Network] -- The PodDisruptionBudget API has been promoted to policy/v1 with no schema changes. The only functional change is that an empty selector (`{}`) written to a policy/v1 PodDisruptionBudget now selects all pods in the namespace. The behavior of the policy/v1beta1 API remains unchanged. The policy/v1beta1 PodDisruptionBudget API is deprecated and will no longer be served in 1.25+. ([#99290](https://github.com/kubernetes/kubernetes/pull/99290), [@mortent](https://github.com/mortent)) [SIG API Machinery, Apps, Auth, Autoscaling, CLI, Cloud Provider, Cluster Lifecycle, Instrumentation, Scheduling and Testing] -- Topology Aware Hints are now available in alpha and can be enabled with the `TopologyAwareHints` feature gate. ([#99522](https://github.com/kubernetes/kubernetes/pull/99522), [@robscott](https://github.com/robscott)) [SIG API Machinery, Apps, Auth, Instrumentation, Network and Testing] - -### Feature - -- Add e2e test to validate performance metrics of volume lifecycle operations ([#94334](https://github.com/kubernetes/kubernetes/pull/94334), [@RaunakShah](https://github.com/RaunakShah)) [SIG Storage and Testing] -- EmptyDir memory backed volumes are sized as the the minimum of pod allocatable memory on a host and an optional explicit user provided value. ([#100319](https://github.com/kubernetes/kubernetes/pull/100319), [@derekwaynecarr](https://github.com/derekwaynecarr)) [SIG Node] -- Enables Kubelet to check volume condition and log events to corresponding pods. ([#99284](https://github.com/kubernetes/kubernetes/pull/99284), [@fengzixu](https://github.com/fengzixu)) [SIG Apps, Instrumentation, Node and Storage] -- Introduce a churn operator to scheduler perf testing framework. ([#98900](https://github.com/kubernetes/kubernetes/pull/98900), [@Huang-Wei](https://github.com/Huang-Wei)) [SIG Scheduling and Testing] -- Kubernetes is now built with Golang 1.16.1 ([#100106](https://github.com/kubernetes/kubernetes/pull/100106), [@justaugustus](https://github.com/justaugustus)) [SIG Cloud Provider, Instrumentation, Release and Testing] -- Migrated pkg/kubelet/cm/devicemanager to structured logging ([#99976](https://github.com/kubernetes/kubernetes/pull/99976), [@knabben](https://github.com/knabben)) [SIG Instrumentation and Node] -- Migrated pkg/kubelet/cm/memorymanager to structured logging ([#99974](https://github.com/kubernetes/kubernetes/pull/99974), [@knabben](https://github.com/knabben)) [SIG Instrumentation and Node] -- Migrated pkg/kubelet/cm/topologymanager to structure logging ([#99969](https://github.com/kubernetes/kubernetes/pull/99969), [@knabben](https://github.com/knabben)) [SIG Instrumentation and Node] -- Rename metrics `etcd_object_counts` to `apiserver_storage_object_counts` and mark it as stable. The original `etcd_object_counts` metrics name is marked as "Deprecated" and will be removed in the future. ([#99785](https://github.com/kubernetes/kubernetes/pull/99785), [@erain](https://github.com/erain)) [SIG API Machinery, Instrumentation and Testing] -- Update pause container to run as pseudo user and group `65535:65535`. This implies the release of version 3.5 of the container images. ([#97963](https://github.com/kubernetes/kubernetes/pull/97963), [@saschagrunert](https://github.com/saschagrunert)) [SIG CLI, Cloud Provider, Cluster Lifecycle, Node, Release, Security and Testing] -- Users might specify the `kubectl.kubernetes.io/default-exec-container` annotation in a Pod to preselect container for kubectl commands. ([#99833](https://github.com/kubernetes/kubernetes/pull/99833), [@mengjiao-liu](https://github.com/mengjiao-liu)) [SIG CLI] - -### Bug or Regression - -- Add ability to skip OpenAPI handler installation to the GenericAPIServer ([#100341](https://github.com/kubernetes/kubernetes/pull/100341), [@kevindelgado](https://github.com/kevindelgado)) [SIG API Machinery] -- Count pod overhead against an entity's ResourceQuota ([#99600](https://github.com/kubernetes/kubernetes/pull/99600), [@gjkim42](https://github.com/gjkim42)) [SIG API Machinery and Node] -- EndpointSlice controllers are less likely to create duplicate EndpointSlices. ([#100103](https://github.com/kubernetes/kubernetes/pull/100103), [@robscott](https://github.com/robscott)) [SIG Apps and Network] -- Ensure only one LoadBalancer rule is created when HA mode is enabled ([#99825](https://github.com/kubernetes/kubernetes/pull/99825), [@feiskyer](https://github.com/feiskyer)) [SIG Cloud Provider] -- Fixed a race condition on API server startup ensuring previously created webhook configurations are effective before the first write request is admitted. ([#95783](https://github.com/kubernetes/kubernetes/pull/95783), [@roycaihw](https://github.com/roycaihw)) [SIG API Machinery] -- Fixed authentication_duration_seconds metric. Previously it included whole apiserver request duration. ([#99944](https://github.com/kubernetes/kubernetes/pull/99944), [@marseel](https://github.com/marseel)) [SIG API Machinery, Instrumentation and Scalability] -- Fixes issue where inline AzueFile secrets could not be accessed from the pod's namespace. ([#100563](https://github.com/kubernetes/kubernetes/pull/100563), [@msau42](https://github.com/msau42)) [SIG Storage] -- Improve speed of vSphere PV provisioning and reduce number of API calls ([#100054](https://github.com/kubernetes/kubernetes/pull/100054), [@gnufied](https://github.com/gnufied)) [SIG Cloud Provider and Storage] -- Kubectl: Fixed panic when describing an ingress backend without an API Group ([#100505](https://github.com/kubernetes/kubernetes/pull/100505), [@lauchokyip](https://github.com/lauchokyip)) [SIG CLI] -- Kubectl: fix case of age column in describe node (#96963, @bl-ue) ([#96963](https://github.com/kubernetes/kubernetes/pull/96963), [@bl-ue](https://github.com/bl-ue)) [SIG CLI] -- Kubelet.exe on Windows now checks that the process running as administrator and the executing user account is listed in the built-in administrators group. This is the equivalent to checking the process is running as uid 0. ([#96616](https://github.com/kubernetes/kubernetes/pull/96616), [@perithompson](https://github.com/perithompson)) [SIG Node and Windows] -- Kubelet: Fixed the bug of getting the number of cpu when the number of cpu logical processors is more than 64 in windows ([#97378](https://github.com/kubernetes/kubernetes/pull/97378), [@hwdef](https://github.com/hwdef)) [SIG Node and Windows] -- Pass `KUBE_BUILD_CONFORMANCE=y` to the package-tarballs to reenable building the conformance tarballs. ([#100571](https://github.com/kubernetes/kubernetes/pull/100571), [@puerco](https://github.com/puerco)) [SIG Release] -- Pod Log stats for windows now reports metrics ([#99221](https://github.com/kubernetes/kubernetes/pull/99221), [@jsturtevant](https://github.com/jsturtevant)) [SIG Node, Storage, Testing and Windows] - -### Other (Cleanup or Flake) - -- A new storage E2E testsuite covers CSIStorageCapacity publishing if a driver opts into the test. ([#100537](https://github.com/kubernetes/kubernetes/pull/100537), [@pohly](https://github.com/pohly)) [SIG Storage and Testing] -- Convert cmd/kubelet/app/server.go to structured logging ([#98334](https://github.com/kubernetes/kubernetes/pull/98334), [@wawa0210](https://github.com/wawa0210)) [SIG Node] -- If kube-apiserver enabled goaway feature, clients required golang 1.15.8 or 1.16+ version to avoid un-expected data race issue. ([#98809](https://github.com/kubernetes/kubernetes/pull/98809), [@answer1991](https://github.com/answer1991)) [SIG API Machinery] -- Increased CSINodeIDMaxLength from 128 bytes to 192 bytes. ([#98753](https://github.com/kubernetes/kubernetes/pull/98753), [@Jiawei0227](https://github.com/Jiawei0227)) [SIG Apps and Storage] -- Migrate `pkg/kubelet/pluginmanager` to structured logging ([#99885](https://github.com/kubernetes/kubernetes/pull/99885), [@qingwave](https://github.com/qingwave)) [SIG Node] -- Migrate `pkg/kubelet/preemption/preemption.go` and `pkg/kubelet/logs/container_log_manager.go` to structured logging ([#99848](https://github.com/kubernetes/kubernetes/pull/99848), [@qingwave](https://github.com/qingwave)) [SIG Node] -- Migrate `pkg/kubelet/(cri)` to structured logging ([#99006](https://github.com/kubernetes/kubernetes/pull/99006), [@yangjunmyfm192085](https://github.com/yangjunmyfm192085)) [SIG Node] -- Migrate `pkg/kubelet/(node, pod)` to structured logging ([#98847](https://github.com/kubernetes/kubernetes/pull/98847), [@yangjunmyfm192085](https://github.com/yangjunmyfm192085)) [SIG Node] -- Migrate `pkg/kubelet/(volume,container)` to structured logging ([#98850](https://github.com/kubernetes/kubernetes/pull/98850), [@yangjunmyfm192085](https://github.com/yangjunmyfm192085)) [SIG Node] -- Migrate `pkg/kubelet/kubelet_node_status.go` to structured logging ([#98154](https://github.com/kubernetes/kubernetes/pull/98154), [@yangjunmyfm192085](https://github.com/yangjunmyfm192085)) [SIG Node and Release] -- Migrate `pkg/kubelet/lifecycle,oom` to structured logging ([#99479](https://github.com/kubernetes/kubernetes/pull/99479), [@mengjiao-liu](https://github.com/mengjiao-liu)) [SIG Instrumentation and Node] -- Migrate cmd/kubelet/+ pkg/kubelet/cadvisor/cadvisor_linux.go + pkg/kubelet/cri/remote/util/util_unix.go + pkg/kubelet/images/image_manager.go to structured logging ([#99994](https://github.com/kubernetes/kubernetes/pull/99994), [@AfrouzMashayekhi](https://github.com/AfrouzMashayekhi)) [SIG Instrumentation and Node] -- Migrate pkg/kubelet/cm/container_manager_linux.go and pkg/kubelet/cm/container_manager_stub.go to structured logging ([#100001](https://github.com/kubernetes/kubernetes/pull/100001), [@shiyajuan123](https://github.com/shiyajuan123)) [SIG Instrumentation and Node] -- Migrate pkg/kubelet/cm/cpumanage/{topology/togit pology.go, policy_none.go, cpu_assignment.go} to structured logging ([#100163](https://github.com/kubernetes/kubernetes/pull/100163), [@lala123912](https://github.com/lala123912)) [SIG Instrumentation and Node] -- Migrate pkg/kubelet/cm/cpumanager/state to structured logging ([#99563](https://github.com/kubernetes/kubernetes/pull/99563), [@jmguzik](https://github.com/jmguzik)) [SIG Instrumentation and Node] -- Migrate pkg/kubelet/config to structured logging ([#100002](https://github.com/kubernetes/kubernetes/pull/100002), [@AfrouzMashayekhi](https://github.com/AfrouzMashayekhi)) [SIG Instrumentation and Node] -- Migrate pkg/kubelet/kubelet.go to structured logging ([#99861](https://github.com/kubernetes/kubernetes/pull/99861), [@navidshaikh](https://github.com/navidshaikh)) [SIG Instrumentation and Node] -- Migrate pkg/kubelet/kubeletconfig to structured logging ([#100265](https://github.com/kubernetes/kubernetes/pull/100265), [@ehashman](https://github.com/ehashman)) [SIG Node] -- Migrate pkg/kubelet/kuberuntime to structured logging ([#99970](https://github.com/kubernetes/kubernetes/pull/99970), [@krzysiekg](https://github.com/krzysiekg)) [SIG Instrumentation and Node] -- Migrate pkg/kubelet/prober to structured logging ([#99830](https://github.com/kubernetes/kubernetes/pull/99830), [@krzysiekg](https://github.com/krzysiekg)) [SIG Instrumentation and Node] -- Migrate pkg/kubelet/winstats to structured logging ([#99855](https://github.com/kubernetes/kubernetes/pull/99855), [@hexxdump](https://github.com/hexxdump)) [SIG Instrumentation and Node] -- Migrate probe log messages to structured logging ([#97093](https://github.com/kubernetes/kubernetes/pull/97093), [@aldudko](https://github.com/aldudko)) [SIG Instrumentation and Node] -- Migrate remaining kubelet files to structured logging ([#100196](https://github.com/kubernetes/kubernetes/pull/100196), [@ehashman](https://github.com/ehashman)) [SIG Instrumentation and Node] -- `apiserver_storage_objects` (a newer version of `etcd_object_counts) is promoted and marked as stable. ([#100082](https://github.com/kubernetes/kubernetes/pull/100082), [@logicalhan](https://github.com/logicalhan)) [SIG API Machinery, Instrumentation and Testing] - -## Dependencies - -### Added -_Nothing has changed._ - -### Changed -- github.com/cilium/ebpf: [1c8d4c9 → v0.2.0](https://github.com/cilium/ebpf/compare/1c8d4c9...v0.2.0) -- github.com/containerd/console: [v1.0.0 → v1.0.1](https://github.com/containerd/console/compare/v1.0.0...v1.0.1) -- github.com/containerd/containerd: [v1.4.1 → v1.4.4](https://github.com/containerd/containerd/compare/v1.4.1...v1.4.4) -- github.com/creack/pty: [v1.1.9 → v1.1.11](https://github.com/creack/pty/compare/v1.1.9...v1.1.11) -- github.com/docker/docker: [bd33bbf → v20.10.2+incompatible](https://github.com/docker/docker/compare/bd33bbf...v20.10.2) -- github.com/google/cadvisor: [v0.38.8 → v0.39.0](https://github.com/google/cadvisor/compare/v0.38.8...v0.39.0) -- github.com/konsorten/go-windows-terminal-sequences: [v1.0.3 → v1.0.2](https://github.com/konsorten/go-windows-terminal-sequences/compare/v1.0.3...v1.0.2) -- github.com/moby/sys/mountinfo: [v0.1.3 → v0.4.0](https://github.com/moby/sys/mountinfo/compare/v0.1.3...v0.4.0) -- github.com/moby/term: [672ec06 → df9cb8a](https://github.com/moby/term/compare/672ec06...df9cb8a) -- github.com/mrunalp/fileutils: [abd8a0e → v0.5.0](https://github.com/mrunalp/fileutils/compare/abd8a0e...v0.5.0) -- github.com/opencontainers/runc: [v1.0.0-rc92 → v1.0.0-rc93](https://github.com/opencontainers/runc/compare/v1.0.0-rc92...v1.0.0-rc93) -- github.com/opencontainers/runtime-spec: [4d89ac9 → e6143ca](https://github.com/opencontainers/runtime-spec/compare/4d89ac9...e6143ca) -- github.com/opencontainers/selinux: [v1.6.0 → v1.8.0](https://github.com/opencontainers/selinux/compare/v1.6.0...v1.8.0) -- github.com/sirupsen/logrus: [v1.6.0 → v1.7.0](https://github.com/sirupsen/logrus/compare/v1.6.0...v1.7.0) -- github.com/syndtr/gocapability: [d983527 → 42c35b4](https://github.com/syndtr/gocapability/compare/d983527...42c35b4) -- github.com/willf/bitset: [d5bec33 → v1.1.11](https://github.com/willf/bitset/compare/d5bec33...v1.1.11) -- gotest.tools/v3: v3.0.2 → v3.0.3 -- k8s.io/klog/v2: v2.5.0 → v2.8.0 -- sigs.k8s.io/structured-merge-diff/v4: v4.0.3 → v4.1.0 - -### Removed -_Nothing has changed._ - - - -# v1.21.0-beta.1 - - -## Downloads for v1.21.0-beta.1 - -### Source Code - -filename | sha512 hash --------- | ----------- -[kubernetes.tar.gz](https://dl.k8s.io/v1.21.0-beta.1/kubernetes.tar.gz) | c9f4f25242e319e5d90f49d26f239a930aad69677c0f3c2387c56bb13482648a26ed234be2bfe2352508f35010e3eb6d3b127c31a9f24fa1e53ac99c38520fe4 -[kubernetes-src.tar.gz](https://dl.k8s.io/v1.21.0-beta.1/kubernetes-src.tar.gz) | 255357db8fa160cab2187658906b674a8b0d9b9a5b5f688cc7b69dc124f5da00362c6cc18ae9b80f7ddb3da6f64c2ab2f12fb9b63a4e063c7366a5375b175cda - -### Client binaries - -filename | sha512 hash --------- | ----------- -[kubernetes-client-darwin-amd64.tar.gz](https://dl.k8s.io/v1.21.0-beta.1/kubernetes-client-darwin-amd64.tar.gz) | 02efd389c8126456416fd2c7ea25c3cc30f612649ad91f631f068d6c0e5e539484d3763cb9a8645ad6b8077e4fcd1552a659d7516ebc4ce6828cf823b65c3016 -[kubernetes-client-darwin-arm64.tar.gz](https://dl.k8s.io/v1.21.0-beta.1/kubernetes-client-darwin-arm64.tar.gz) | ac90dcd1699d1d7ff9c8342d481f6d0d97ccdc3ec501a56dc7c9e1898a8f77f712bf66942d304bfe581b5494f13e3efa211865de88f89749780e9e26e673dbdb -[kubernetes-client-linux-386.tar.gz](https://dl.k8s.io/v1.21.0-beta.1/kubernetes-client-linux-386.tar.gz) | cce5fb84cc7a1ee664f89d8ad3064307c51c044e9ddd2ae5a004939b69d3b3ef6f29acc5782e27d0c8f0d6d3d9c96e922f5d1b99d210ca3e754666d775df9f0c -[kubernetes-client-linux-amd64.tar.gz](https://dl.k8s.io/v1.21.0-beta.1/kubernetes-client-linux-amd64.tar.gz) | 2e93bbd2e60ad7cd8fe495115e96c55b1dc8facd100a827ef9c197a732679b60cceb9ea7bf92a1f5e328c3b8adfa8d3922cbc5d8370e374f3381b83f5b877b4f -[kubernetes-client-linux-arm.tar.gz](https://dl.k8s.io/v1.21.0-beta.1/kubernetes-client-linux-arm.tar.gz) | 23f03b6a8fa9decce9b89a2c1bd3dae6d0b2f9e533e35a79e2c5a29326a165259677594ae83c877219a21bdb95557a284e55f4eec12954742794579c89a7d7e5 -[kubernetes-client-linux-arm64.tar.gz](https://dl.k8s.io/v1.21.0-beta.1/kubernetes-client-linux-arm64.tar.gz) | 3acf3101b46568b0ded6b90f13df0e918870d6812dc1a584903ddb8ba146484a204b9e442f863df47c7d4dab043fd9f7294c5510d3eb09004993d6d3b1e9e13c -[kubernetes-client-linux-ppc64le.tar.gz](https://dl.k8s.io/v1.21.0-beta.1/kubernetes-client-linux-ppc64le.tar.gz) | f749198df69577f62872d3096138a1b8969ec6b1636eb68eb56640bf33cf5f97a11df4363462749a1c0dc3ccbb8ae76c5d66864bf1c5cf7e52599caaf498e504 -[kubernetes-client-linux-s390x.tar.gz](https://dl.k8s.io/v1.21.0-beta.1/kubernetes-client-linux-s390x.tar.gz) | 3f6c0189d59fca22cdded3a02c672ef703d17e6ab0831e173a870e14ccec436c142600e9fc35b403571b6906f2be8d18d38d33330f7caada971bbe1187b388f6 -[kubernetes-client-windows-386.tar.gz](https://dl.k8s.io/v1.21.0-beta.1/kubernetes-client-windows-386.tar.gz) | 03d92371c425cf331c80807c0ac56f953be304fc6719057258a363d527d186d610e1d4b4d401b34128062983265c2e21f2d2389231aa66a6f5787eee78142cf6 -[kubernetes-client-windows-amd64.tar.gz](https://dl.k8s.io/v1.21.0-beta.1/kubernetes-client-windows-amd64.tar.gz) | 489ece0c886a025ca3a25d28518637a5a824ea6544e7ef8778321036f13c8909a978ad4ceca966cec1e1cda99f25ca78bfd37460d1231c77436d216d43c872ad - -### Server binaries - -filename | sha512 hash --------- | ----------- -[kubernetes-server-linux-amd64.tar.gz](https://dl.k8s.io/v1.21.0-beta.1/kubernetes-server-linux-amd64.tar.gz) | 2e95cb31d5afcb6842c41d25b7d0c18dd7e65693b2d93c8aa44e5275f9c6201e1a67685c7a8ddefa334babb04cb559d26e39b6a18497695a07dc270568cae108 -[kubernetes-server-linux-arm.tar.gz](https://dl.k8s.io/v1.21.0-beta.1/kubernetes-server-linux-arm.tar.gz) | 2927e82b98404c077196ce3968f3afd51a7576aa56d516019bd3976771c0213ba01e78da5b77478528e770da0d334e9457995fafb98820ed68b2ee34beb68856 -[kubernetes-server-linux-arm64.tar.gz](https://dl.k8s.io/v1.21.0-beta.1/kubernetes-server-linux-arm64.tar.gz) | e0f7aea3ea598214a9817bc04949389cb7e4e7b9503141a590ef48c0b681fe44a4243ebc6280752fa41aa1093149b3ee1bcef7664edb746097a342281825430b -[kubernetes-server-linux-ppc64le.tar.gz](https://dl.k8s.io/v1.21.0-beta.1/kubernetes-server-linux-ppc64le.tar.gz) | c011f7eb01294e9ba5d5ced719068466f88ed595dcb8d554a36a4dd5118fb6b3d6bafe8bf89aa2d42988e69793ed777ba77b8876c6ec74f898a43cfce1f61bf4 -[kubernetes-server-linux-s390x.tar.gz](https://dl.k8s.io/v1.21.0-beta.1/kubernetes-server-linux-s390x.tar.gz) | 15f6683e7f16caab7eebead2b7c15799460abbf035a43de0b75f96b0be19908f58add98a777a0cca916230d60cf6bfe3fee92b9dcff50274b1e37c243c157969 - -### Node binaries - -filename | sha512 hash --------- | ----------- -[kubernetes-node-linux-amd64.tar.gz](https://dl.k8s.io/v1.21.0-beta.1/kubernetes-node-linux-amd64.tar.gz) | ed58679561197110f366b9109f7afd62c227bfc271918ccf3eea203bb2ab6428eb5db4dd6c965f202a8a636f66da199470269b863815809b99d53d2fa47af2ea -[kubernetes-node-linux-arm.tar.gz](https://dl.k8s.io/v1.21.0-beta.1/kubernetes-node-linux-arm.tar.gz) | 7e6c7f1957fcdecec8fef689c5019edbc0d0c11d22dafbfef0a07121d10d8f6273644f73511bd06a9a88b04d81a940bd6645ffb5711422af64af547a45c76273 -[kubernetes-node-linux-arm64.tar.gz](https://dl.k8s.io/v1.21.0-beta.1/kubernetes-node-linux-arm64.tar.gz) | a3618f29967e7a1574917a67f0296e65780321eda484b99aa32bfd4dc9b35acdefce33da952ac52dfb509fbac5bf700cf177431fad2ab4adcab0544538939faa -[kubernetes-node-linux-ppc64le.tar.gz](https://dl.k8s.io/v1.21.0-beta.1/kubernetes-node-linux-ppc64le.tar.gz) | 326d3eb521b41bdf489912177f70b8cdd7cd828bb9b3d847ed3694eb27e457f24e0a88b8e51b726eee39800a3c5a40c1b30e3a8ec4a34d8041b3d8ef05d1b749 -[kubernetes-node-linux-s390x.tar.gz](https://dl.k8s.io/v1.21.0-beta.1/kubernetes-node-linux-s390x.tar.gz) | 022d05ebaa66a0332c4fe18cdaf23d14c2c7e4d1f2af7f27baaf1eb042e6890dc3434b4ac8ba58c35d590717956f8c3458112685aff4938b94b18e263c3f4256 -[kubernetes-node-windows-amd64.tar.gz](https://dl.k8s.io/v1.21.0-beta.1/kubernetes-node-windows-amd64.tar.gz) | fa691ed93f07af6bc1cf57e20a30580d6c528f88e5fea3c14f39c1820969dc5a0eb476c5b87b288593d0c086c4dd93aff6165082393283c3f46c210f9bb66d61 - -## Changelog since v1.21.0-beta.0 - -## Urgent Upgrade Notes - -### (No, really, you MUST read this before you upgrade) - - - Kubeadm: during "init" an empty cgroupDriver value in the KubeletConfiguration is now always set to "systemd" unless the user is explicit about it. This requires existing machine setups to configure the container runtime to use the "systemd" driver. Documentation on this topic can be found here: https://kubernetes.io/docs/setup/production-environment/container-runtimes/. When upgrading existing clusters / nodes using "kubeadm upgrade" the old cgroupDriver value is preserved, but in 1.22 this change will also apply to "upgrade". For more information on migrating to the "systemd" driver or remaining on the "cgroupfs" driver see: https://kubernetes.io/docs/tasks/administer-cluster/kubeadm/configure-cgroup-driver/. ([#99471](https://github.com/kubernetes/kubernetes/pull/99471), [@neolit123](https://github.com/neolit123)) [SIG Cluster Lifecycle] - - Migrate `pkg/kubelet/(dockershim, network)` to structured logging - Exit code changed from 255 to 1 ([#98939](https://github.com/kubernetes/kubernetes/pull/98939), [@yangjunmyfm192085](https://github.com/yangjunmyfm192085)) [SIG Network and Node] - - Migrate `pkg/kubelet/certificate` to structured logging - Exit code changed from 255 to 1 ([#98993](https://github.com/kubernetes/kubernetes/pull/98993), [@SataQiu](https://github.com/SataQiu)) [SIG Auth and Node] - - Newly provisioned PVs by EBS plugin will no longer use the deprecated "failure-domain.beta.kubernetes.io/zone" and "failure-domain.beta.kubernetes.io/region" labels. It will use "topology.kubernetes.io/zone" and "topology.kubernetes.io/region" labels instead. ([#99130](https://github.com/kubernetes/kubernetes/pull/99130), [@ayberk](https://github.com/ayberk)) [SIG Cloud Provider, Storage and Testing] - - Newly provisioned PVs by OpenStack Cinder plugin will no longer use the deprecated "failure-domain.beta.kubernetes.io/zone" and "failure-domain.beta.kubernetes.io/region" labels. It will use "topology.kubernetes.io/zone" and "topology.kubernetes.io/region" labels instead. ([#99719](https://github.com/kubernetes/kubernetes/pull/99719), [@jsafrane](https://github.com/jsafrane)) [SIG Cloud Provider and Storage] - - OpenStack Cinder CSI migration is on by default, Clinder CSI driver must be installed on clusters on OpenStack for Cinder volumes to work. ([#98538](https://github.com/kubernetes/kubernetes/pull/98538), [@dims](https://github.com/dims)) [SIG Storage] - - Package pkg/kubelet/server migrated to structured logging - Exit code changed from 255 to 1 ([#99838](https://github.com/kubernetes/kubernetes/pull/99838), [@adisky](https://github.com/adisky)) [SIG Node] - - Pkg/kubelet/kuberuntime/kuberuntime_manager.go migrated to structured logging - Exit code changed from 255 to 1 ([#99841](https://github.com/kubernetes/kubernetes/pull/99841), [@adisky](https://github.com/adisky)) [SIG Instrumentation and Node] - -## Changes by Kind - -### Deprecation - -- Kubeadm: the deprecated kube-dns is no longer supported as an option. If "ClusterConfiguration.dns.type" is set to "kube-dns" kubeadm will now throw an error. ([#99646](https://github.com/kubernetes/kubernetes/pull/99646), [@rajansandeep](https://github.com/rajansandeep)) [SIG Cluster Lifecycle] -- Remove deprecated --generator --replicas --service-generator --service-overrides --schedule from kubectl run - Deprecate --serviceaccount --hostport --requests --limits in kubectl run ([#99732](https://github.com/kubernetes/kubernetes/pull/99732), [@soltysh](https://github.com/soltysh)) [SIG CLI and Testing] -- `audit.k8s.io/v1beta1` and `audit.k8s.io/v1alpha1` audit policy configuration and audit events are deprecated in favor of `audit.k8s.io/v1`, available since v1.13. kube-apiserver invocations that specify alpha or beta policy configurations with `--audit-policy-file`, or explicitly request alpha or beta audit events with `--audit-log-version` / `--audit-webhook-version` must update to use `audit.k8s.io/v1` and accept `audit.k8s.io/v1` events prior to v1.24. ([#98858](https://github.com/kubernetes/kubernetes/pull/98858), [@carlory](https://github.com/carlory)) [SIG Auth] -- `diskformat` stroage class parameter for in-tree vSphere volume plugin is deprecated as of v1.21 release. Please consider updating storageclass and remove `diskformat` parameter. vSphere CSI Driver does not support diskformat storageclass parameter. - - vSphere releases less than 67u3 are deprecated as of v1.21. Please consider upgrading vSphere to 67u3 or above. vSphere CSI Driver requires minimum vSphere 67u3. - - VM Hardware version less than 15 is deprecated as of v1.21. Please consider upgrading the Node VM Hardware version to 15 or above. vSphere CSI Driver recommends Node VM's Hardware version set to at least vmx-15. - - Multi vCenter support is deprecated as of v1.21. If you have a Kubernetes cluster spanning across multiple vCenter servers, please consider moving all k8s nodes to a single vCenter Server. vSphere CSI Driver does not support Kubernetes deployment spanning across multiple vCenter servers. - - Support for these deprecations will be available till Kubernetes v1.24. ([#98546](https://github.com/kubernetes/kubernetes/pull/98546), [@divyenpatel](https://github.com/divyenpatel)) [SIG Cloud Provider and Storage] - -### API Change - -- 1. PodAffinityTerm includes a namespaceSelector field to allow selecting eligible namespaces based on their labels. - 2. A new CrossNamespacePodAffinity quota scope API that allows restricting which namespaces allowed to use PodAffinityTerm with corss-namespace reference via namespaceSelector or namespaces fields. ([#98582](https://github.com/kubernetes/kubernetes/pull/98582), [@ahg-g](https://github.com/ahg-g)) [SIG API Machinery, Apps, Auth and Testing] -- Add a default metadata name labels for selecting any namespace by its name. ([#96968](https://github.com/kubernetes/kubernetes/pull/96968), [@jayunit100](https://github.com/jayunit100)) [SIG API Machinery, Apps, Cloud Provider, Storage and Testing] -- Added `.spec.completionMode` field to Job, with accepted values `NonIndexed` (default) and `Indexed` ([#98441](https://github.com/kubernetes/kubernetes/pull/98441), [@alculquicondor](https://github.com/alculquicondor)) [SIG Apps and CLI] -- Clarified NetworkPolicy policyTypes documentation ([#97216](https://github.com/kubernetes/kubernetes/pull/97216), [@joejulian](https://github.com/joejulian)) [SIG Network] -- DaemonSets accept a MaxSurge integer or percent on their rolling update strategy that will launch the updated pod on nodes and wait for those pods to go ready before marking the old out-of-date pods as deleted. This allows workloads to avoid downtime during upgrades when deployed using DaemonSets. This feature is alpha and is behind the DaemonSetUpdateSurge feature gate. ([#96441](https://github.com/kubernetes/kubernetes/pull/96441), [@smarterclayton](https://github.com/smarterclayton)) [SIG Apps and Testing] -- EndpointSlice API is now GA. The EndpointSlice topology field has been removed from the GA API and will be replaced by a new per Endpoint Zone field. If the topology field was previously used, it will be converted into an annotation in the v1 Resource. The discovery.k8s.io/v1alpha1 API is removed. ([#99662](https://github.com/kubernetes/kubernetes/pull/99662), [@swetharepakula](https://github.com/swetharepakula)) [SIG API Machinery, CLI, Cloud Provider, Cluster Lifecycle, Instrumentation, Network and Testing] -- EndpointSlice Controllers are now GA. The EndpointSlice Controller will not populate the `deprecatedTopology` field and will only provide topology information through the `zone` and `nodeName` fields. ([#99870](https://github.com/kubernetes/kubernetes/pull/99870), [@swetharepakula](https://github.com/swetharepakula)) [SIG API Machinery, Apps, Auth, Network and Testing] -- IngressClass resource can now reference a resource in a specific namespace - for implementation-specific configuration(previously only Cluster-level resources were allowed). - This feature can be enabled using the IngressClassNamespacedParams feature gate. ([#99275](https://github.com/kubernetes/kubernetes/pull/99275), [@hbagdi](https://github.com/hbagdi)) [SIG API Machinery, CLI and Network] -- Introduce conditions for PodDisruptionBudget ([#98127](https://github.com/kubernetes/kubernetes/pull/98127), [@mortent](https://github.com/mortent)) [SIG API Machinery, Apps, Auth, CLI, Cloud Provider, Cluster Lifecycle and Instrumentation] -- Jobs API has a new .spec.suspend field that can be used to suspend and resume Jobs ([#98727](https://github.com/kubernetes/kubernetes/pull/98727), [@adtac](https://github.com/adtac)) [SIG API Machinery, Apps, Node, Scheduling and Testing] -- Kubelet Graceful Node Shutdown feature is now beta. ([#99735](https://github.com/kubernetes/kubernetes/pull/99735), [@bobbypage](https://github.com/bobbypage)) [SIG Node] -- Limit the quest value of hugepage to integer multiple of page size. ([#98515](https://github.com/kubernetes/kubernetes/pull/98515), [@lala123912](https://github.com/lala123912)) [SIG Apps] -- One new field "InternalTrafficPolicy" in Service is added. - It specifies if the cluster internal traffic should be routed to all endpoints or node-local endpoints only. - "Cluster" routes internal traffic to a Service to all endpoints. - "Local" routes traffic to node-local endpoints only, and traffic is dropped if no node-local endpoints are ready. - The default value is "Cluster". ([#96600](https://github.com/kubernetes/kubernetes/pull/96600), [@maplain](https://github.com/maplain)) [SIG API Machinery, Apps and Network] -- PodSecurityPolicy only stores "generic" as allowed volume type if the GenericEphemeralVolume feature gate is enabled ([#98918](https://github.com/kubernetes/kubernetes/pull/98918), [@pohly](https://github.com/pohly)) [SIG Auth and Security] -- Promote CronJobs to batch/v1 ([#99423](https://github.com/kubernetes/kubernetes/pull/99423), [@soltysh](https://github.com/soltysh)) [SIG API Machinery, Apps, CLI and Testing] -- Remove support for building Kubernetes with bazel. ([#99561](https://github.com/kubernetes/kubernetes/pull/99561), [@BenTheElder](https://github.com/BenTheElder)) [SIG API Machinery, Apps, Architecture, Auth, Autoscaling, CLI, Cloud Provider, Cluster Lifecycle, Instrumentation, Network, Node, Release, Scalability, Scheduling, Storage, Testing and Windows] -- Setting loadBalancerClass in load balancer type of service is available with this PR. - Users who want to use a custom load balancer can specify loadBalancerClass to achieve it. ([#98277](https://github.com/kubernetes/kubernetes/pull/98277), [@XudongLiuHarold](https://github.com/XudongLiuHarold)) [SIG API Machinery, Apps, Cloud Provider and Network] -- Storage capacity tracking (= the CSIStorageCapacity feature) is beta, storage.k8s.io/v1alpha1/VolumeAttachment and storage.k8s.io/v1alpha1/CSIStorageCapacity objects are deprecated ([#99641](https://github.com/kubernetes/kubernetes/pull/99641), [@pohly](https://github.com/pohly)) [SIG API Machinery, Apps, Auth, Scheduling, Storage and Testing] -- Support for Indexed Job: a Job that is considered completed when Pods associated to indexes from 0 to (.spec.completions-1) have succeeded. ([#98812](https://github.com/kubernetes/kubernetes/pull/98812), [@alculquicondor](https://github.com/alculquicondor)) [SIG Apps and CLI] -- The apiserver now resets managedFields that got corrupted by a mutating admission controller. ([#98074](https://github.com/kubernetes/kubernetes/pull/98074), [@kwiesmueller](https://github.com/kwiesmueller)) [SIG API Machinery and Testing] -- `controller.kubernetes.io/pod-deletion-cost` annotation can be set to offer a hint on the cost of deleting a pod compared to other pods belonging to the same ReplicaSet. Pods with lower deletion cost are deleted first. This is an alpha feature. ([#99163](https://github.com/kubernetes/kubernetes/pull/99163), [@ahg-g](https://github.com/ahg-g)) [SIG Apps] - -### Feature - -- A client-go metric, rest_client_exec_plugin_call_total, has been added to track total calls to client-go credential plugins. ([#98892](https://github.com/kubernetes/kubernetes/pull/98892), [@ankeesler](https://github.com/ankeesler)) [SIG API Machinery, Auth, Cluster Lifecycle and Instrumentation] -- Add --use-protocol-buffers flag to kubectl top pods and nodes ([#96655](https://github.com/kubernetes/kubernetes/pull/96655), [@serathius](https://github.com/serathius)) [SIG CLI] -- Add support to generate client-side binaries for new darwin/arm64 platform ([#97743](https://github.com/kubernetes/kubernetes/pull/97743), [@dims](https://github.com/dims)) [SIG Release and Testing] -- Added `ephemeral_volume_controller_create[_failures]_total` counters to kube-controller-manager metrics ([#99115](https://github.com/kubernetes/kubernetes/pull/99115), [@pohly](https://github.com/pohly)) [SIG API Machinery, Apps, Cluster Lifecycle, Instrumentation and Storage] -- Adds alpha feature `VolumeCapacityPriority` which makes the scheduler prioritize nodes based on the best matching size of statically provisioned PVs across multiple topologies. ([#96347](https://github.com/kubernetes/kubernetes/pull/96347), [@cofyc](https://github.com/cofyc)) [SIG Apps, Network, Scheduling, Storage and Testing] -- Adds two new metrics to cronjobs, a histogram to track the time difference when a job is created and the expected time when it should be created, and a gauge for the missed schedules of a cronjob ([#99341](https://github.com/kubernetes/kubernetes/pull/99341), [@alaypatel07](https://github.com/alaypatel07)) [SIG Apps and Instrumentation] -- Alpha implementation of Kubectl Command Headers: SIG CLI KEP 859 enabled when KUBECTL_COMMAND_HEADERS environment variable set on the client command line. - - To enable: export KUBECTL_COMMAND_HEADERS=1; kubectl ... ([#98952](https://github.com/kubernetes/kubernetes/pull/98952), [@seans3](https://github.com/seans3)) [SIG API Machinery and CLI] -- Component owner can configure the allowlist of metric label with flag '--allow-metric-labels'. ([#99738](https://github.com/kubernetes/kubernetes/pull/99738), [@YoyinZyc](https://github.com/YoyinZyc)) [SIG API Machinery, Cluster Lifecycle and Instrumentation] -- Disruption controller only sends one event per PodDisruptionBudget if scale can't be computed ([#98128](https://github.com/kubernetes/kubernetes/pull/98128), [@mortent](https://github.com/mortent)) [SIG Apps] -- EndpointSliceNodeName will always be enabled, so NodeName will always be available in the v1beta1 API. ([#99746](https://github.com/kubernetes/kubernetes/pull/99746), [@swetharepakula](https://github.com/swetharepakula)) [SIG Apps and Network] -- Graduate CRIContainerLogRotation feature gate to GA. ([#99651](https://github.com/kubernetes/kubernetes/pull/99651), [@umohnani8](https://github.com/umohnani8)) [SIG Node and Testing] -- Kube-proxy iptables: new metric sync_proxy_rules_iptables_total that exposes the number of rules programmed per table in each iteration ([#99653](https://github.com/kubernetes/kubernetes/pull/99653), [@aojea](https://github.com/aojea)) [SIG Instrumentation and Network] -- Kube-scheduler now logs plugin scoring summaries at --v=4 ([#99411](https://github.com/kubernetes/kubernetes/pull/99411), [@damemi](https://github.com/damemi)) [SIG Scheduling] -- Kubeadm: a warning to user as ipv6 site-local is deprecated ([#99574](https://github.com/kubernetes/kubernetes/pull/99574), [@pacoxu](https://github.com/pacoxu)) [SIG Cluster Lifecycle and Network] -- Kubeadm: apply the "node.kubernetes.io/exclude-from-external-load-balancers" label on control plane nodes during "init", "join" and "upgrade" to preserve backwards compatibility with the lagacy LB mode where nodes labeled as "master" where excluded. To opt-out you can remove the label from a node. See #97543 and the linked KEP for more details. ([#98269](https://github.com/kubernetes/kubernetes/pull/98269), [@neolit123](https://github.com/neolit123)) [SIG Cluster Lifecycle] -- Kubeadm: if the user has customized their image repository via the kubeadm configuration, pass the custom pause image repository and tag to the kubelet via --pod-infra-container-image not only for Docker but for all container runtimes. This flag tells the kubelet that it should not garbage collect the image. ([#99476](https://github.com/kubernetes/kubernetes/pull/99476), [@neolit123](https://github.com/neolit123)) [SIG Cluster Lifecycle] -- Kubeadm: promote IPv6DualStack feature gate to Beta ([#99294](https://github.com/kubernetes/kubernetes/pull/99294), [@pacoxu](https://github.com/pacoxu)) [SIG Cluster Lifecycle] -- Kubectl version changed to write a warning message to stderr if the client and server version difference exceeds the supported version skew of +/-1 minor version. ([#98250](https://github.com/kubernetes/kubernetes/pull/98250), [@brianpursley](https://github.com/brianpursley)) [SIG CLI] -- Kubernetes is now built with Golang 1.16 ([#98572](https://github.com/kubernetes/kubernetes/pull/98572), [@justaugustus](https://github.com/justaugustus)) [SIG API Machinery, Auth, CLI, Cloud Provider, Cluster Lifecycle, Instrumentation, Node, Release and Testing] -- Persistent Volumes formatted with the btrfs filesystem will now automatically resize when expanded. ([#99361](https://github.com/kubernetes/kubernetes/pull/99361), [@Novex](https://github.com/Novex)) [SIG Storage] -- Remove cAdvisor json metrics api collected by Kubelet ([#99236](https://github.com/kubernetes/kubernetes/pull/99236), [@pacoxu](https://github.com/pacoxu)) [SIG Node] -- Sysctls is now GA and locked to default ([#99158](https://github.com/kubernetes/kubernetes/pull/99158), [@wgahnagl](https://github.com/wgahnagl)) [SIG Node] -- The NodeAffinity plugin implements the PreFilter extension, offering enhanced performance for Filter. ([#99213](https://github.com/kubernetes/kubernetes/pull/99213), [@AliceZhang2016](https://github.com/AliceZhang2016)) [SIG Scheduling] -- The endpointslice mirroring controller mirrors endpoints annotations and labels to the generated endpoint slices, it also ensures that updates on any of these fields are mirrored. - The well-known annotation endpoints.kubernetes.io/last-change-trigger-time is skipped and not mirrored. ([#98116](https://github.com/kubernetes/kubernetes/pull/98116), [@aojea](https://github.com/aojea)) [SIG Apps, Network and Testing] -- Update the latest validated version of Docker to 20.10 ([#98977](https://github.com/kubernetes/kubernetes/pull/98977), [@neolit123](https://github.com/neolit123)) [SIG CLI, Cluster Lifecycle and Node] -- Upgrade node local dns to 1.17.0 for better IPv6 support ([#99749](https://github.com/kubernetes/kubernetes/pull/99749), [@pacoxu](https://github.com/pacoxu)) [SIG Cloud Provider and Network] -- Users might specify the `kubectl.kubernetes.io/default-exec-container` annotation in a Pod to preselect container for kubectl commands. ([#99581](https://github.com/kubernetes/kubernetes/pull/99581), [@mengjiao-liu](https://github.com/mengjiao-liu)) [SIG CLI] -- When downscaling ReplicaSets, ready and creation timestamps are compared in a logarithmic scale. ([#99212](https://github.com/kubernetes/kubernetes/pull/99212), [@damemi](https://github.com/damemi)) [SIG Apps and Testing] -- When the kubelet is watching a ConfigMap or Secret purely in the context of setting environment variables - for containers, only hold that watch for a defined duration before cancelling it. This change reduces the CPU - and memory usage of the kube-apiserver in large clusters. ([#99393](https://github.com/kubernetes/kubernetes/pull/99393), [@chenyw1990](https://github.com/chenyw1990)) [SIG API Machinery, Node and Testing] -- WindowsEndpointSliceProxying feature gate has graduated to beta and is enabled by default. This means kube-proxy will read from EndpointSlices instead of Endpoints on Windows by default. ([#99794](https://github.com/kubernetes/kubernetes/pull/99794), [@robscott](https://github.com/robscott)) [SIG Network] - -### Bug or Regression - -- Creating a PVC with DataSource should fail for non-CSI plugins. ([#97086](https://github.com/kubernetes/kubernetes/pull/97086), [@xing-yang](https://github.com/xing-yang)) [SIG Apps and Storage] -- EndpointSlice controller is now less likely to emit FailedToUpdateEndpointSlices events. ([#99345](https://github.com/kubernetes/kubernetes/pull/99345), [@robscott](https://github.com/robscott)) [SIG Apps and Network] -- EndpointSliceMirroring controller is now less likely to emit FailedToUpdateEndpointSlices events. ([#99756](https://github.com/kubernetes/kubernetes/pull/99756), [@robscott](https://github.com/robscott)) [SIG Apps and Network] -- Fix --ignore-errors does not take effect if multiple logs are printed and unfollowed ([#97686](https://github.com/kubernetes/kubernetes/pull/97686), [@wzshiming](https://github.com/wzshiming)) [SIG CLI] -- Fix bug that would let the Horizontal Pod Autoscaler scale down despite at least one metric being unavailable/invalid ([#99514](https://github.com/kubernetes/kubernetes/pull/99514), [@mikkeloscar](https://github.com/mikkeloscar)) [SIG Apps and Autoscaling] -- Fix cgroup handling for systemd with cgroup v2 ([#98365](https://github.com/kubernetes/kubernetes/pull/98365), [@odinuge](https://github.com/odinuge)) [SIG Node] -- Fix smb mount PermissionDenied issue on Windows ([#99550](https://github.com/kubernetes/kubernetes/pull/99550), [@andyzhangx](https://github.com/andyzhangx)) [SIG Cloud Provider, Storage and Windows] -- Fixed a bug that causes smaller number of conntrack-max being used under CPU static policy. (#99225, @xh4n3) ([#99613](https://github.com/kubernetes/kubernetes/pull/99613), [@xh4n3](https://github.com/xh4n3)) [SIG Network] -- Fixed bug that caused cAdvisor to incorrectly detect single-socket multi-NUMA topology. ([#99315](https://github.com/kubernetes/kubernetes/pull/99315), [@iwankgb](https://github.com/iwankgb)) [SIG Node] -- Fixes add-on manager leader election ([#98968](https://github.com/kubernetes/kubernetes/pull/98968), [@liggitt](https://github.com/liggitt)) [SIG Cloud Provider] -- Improved update time of pod statuses following new probe results. ([#98376](https://github.com/kubernetes/kubernetes/pull/98376), [@matthyx](https://github.com/matthyx)) [SIG Node and Testing] -- Kube-apiserver: an update of a pod with a generic ephemeral volume dropped that volume if the feature had been disabled since creating the pod with such a volume ([#99446](https://github.com/kubernetes/kubernetes/pull/99446), [@pohly](https://github.com/pohly)) [SIG Apps, Node and Storage] -- Kubeadm: skip validating pod subnet against node-cidr-mask when allocate-node-cidrs is set to be false ([#98984](https://github.com/kubernetes/kubernetes/pull/98984), [@SataQiu](https://github.com/SataQiu)) [SIG Cluster Lifecycle] -- On single-stack configured (IPv4 or IPv6, but not both) clusters, Services which are both headless (no clusterIP) and selectorless (empty or undefined selector) will report `ipFamilyPolicy RequireDualStack` and will have entries in `ipFamilies[]` for both IPv4 and IPv6. This is a change from alpha, but does not have any impact on the manually-specified Endpoints and EndpointSlices for the Service. ([#99555](https://github.com/kubernetes/kubernetes/pull/99555), [@thockin](https://github.com/thockin)) [SIG Apps and Network] -- Resolves spurious `Failed to list *v1.Secret` or `Failed to list *v1.ConfigMap` messages in kubelet logs. ([#99538](https://github.com/kubernetes/kubernetes/pull/99538), [@liggitt](https://github.com/liggitt)) [SIG Auth and Node] -- Return zero time (midnight on Jan. 1, 1970) instead of negative number when reporting startedAt and finishedAt of the not started or a running Pod when using dockershim as a runtime. ([#99585](https://github.com/kubernetes/kubernetes/pull/99585), [@Iceber](https://github.com/Iceber)) [SIG Node] -- Stdin is now only passed to client-go exec credential plugins when it is detected to be an interactive terminal. Previously, it was passed to client-go exec plugins when **stdout*- was detected to be an interactive terminal. ([#99654](https://github.com/kubernetes/kubernetes/pull/99654), [@ankeesler](https://github.com/ankeesler)) [SIG API Machinery and Auth] -- The maximum number of ports allowed in EndpointSlices has been increased from 100 to 20,000 ([#99795](https://github.com/kubernetes/kubernetes/pull/99795), [@robscott](https://github.com/robscott)) [SIG Network] -- Updates the commands - - kubectl kustomize {arg} - - kubectl apply -k {arg} - to use same code as kustomize CLI v4.0.5 - - [v4.0.5]: https://github.com/kubernetes-sigs/kustomize/releases/tag/kustomize%2Fv4.0.5 ([#98946](https://github.com/kubernetes/kubernetes/pull/98946), [@monopole](https://github.com/monopole)) [SIG API Machinery, Architecture, CLI, Cloud Provider, Cluster Lifecycle, Instrumentation, Node and Storage] -- When a CNI plugin returns dual-stack pod IPs, kubelet will now try to respect the - "primary IP family" of the cluster by picking a primary pod IP of the same family - as the (primary) node IP, rather than assuming that the CNI plugin returned the IPs - in the order the administrator wanted (since some CNI plugins don't allow - configuring this). ([#97979](https://github.com/kubernetes/kubernetes/pull/97979), [@danwinship](https://github.com/danwinship)) [SIG Network and Node] -- When using Containerd on Windows, the "C:\Windows\System32\drivers\etc\hosts" file will now be managed by kubelet. ([#83730](https://github.com/kubernetes/kubernetes/pull/83730), [@claudiubelu](https://github.com/claudiubelu)) [SIG Node and Windows] -- `VolumeBindingArgs` now allow `BindTimeoutSeconds` to be set as zero, while the value zero indicates no waiting for the checking of volume binding operation. ([#99835](https://github.com/kubernetes/kubernetes/pull/99835), [@chendave](https://github.com/chendave)) [SIG Scheduling and Storage] -- `kubectl exec` and `kubectl attach` now honor the `--quiet` flag which suppresses output from the local binary that could be confused by a script with the remote command output (all non-failure output is hidden). In addition, print inline with exec and attach the list of alternate containers when we default to the first spec.container. ([#99004](https://github.com/kubernetes/kubernetes/pull/99004), [@smarterclayton](https://github.com/smarterclayton)) [SIG CLI] - -### Other (Cleanup or Flake) - -- Apiserver_request_duration_seconds is promoted to stable status. ([#99925](https://github.com/kubernetes/kubernetes/pull/99925), [@logicalhan](https://github.com/logicalhan)) [SIG API Machinery, Instrumentation and Testing] -- Apiserver_request_total is promoted to stable status and no longer has a content-type dimensions, so any alerts/charts which presume the existence of this will fail. This is however, unlikely to be the case since it was effectively an unbounded dimension in the first place. ([#99788](https://github.com/kubernetes/kubernetes/pull/99788), [@logicalhan](https://github.com/logicalhan)) [SIG API Machinery, Instrumentation and Testing] -- EndpointSlice generation is now incremented when labels change. ([#99750](https://github.com/kubernetes/kubernetes/pull/99750), [@robscott](https://github.com/robscott)) [SIG Network] -- Featuregate AllowInsecureBackendProxy is promoted to GA ([#99658](https://github.com/kubernetes/kubernetes/pull/99658), [@deads2k](https://github.com/deads2k)) [SIG API Machinery] -- Migrate `pkg/kubelet/(eviction)` to structured logging ([#99032](https://github.com/kubernetes/kubernetes/pull/99032), [@yangjunmyfm192085](https://github.com/yangjunmyfm192085)) [SIG Node] -- Migrate deployment controller log messages to structured logging ([#97507](https://github.com/kubernetes/kubernetes/pull/97507), [@aldudko](https://github.com/aldudko)) [SIG Apps] -- Migrate pkg/kubelet/cloudresource to structured logging ([#98999](https://github.com/kubernetes/kubernetes/pull/98999), [@sladyn98](https://github.com/sladyn98)) [SIG Node] -- Migrate pkg/kubelet/cri/remote logs to structured logging ([#98589](https://github.com/kubernetes/kubernetes/pull/98589), [@chenyw1990](https://github.com/chenyw1990)) [SIG Node] -- Migrate pkg/kubelet/kuberuntime/kuberuntime_container.go logs to structured logging ([#96973](https://github.com/kubernetes/kubernetes/pull/96973), [@chenyw1990](https://github.com/chenyw1990)) [SIG Instrumentation and Node] -- Migrate pkg/kubelet/status to structured logging ([#99836](https://github.com/kubernetes/kubernetes/pull/99836), [@navidshaikh](https://github.com/navidshaikh)) [SIG Instrumentation and Node] -- Migrate pkg/kubelet/token to structured logging ([#99264](https://github.com/kubernetes/kubernetes/pull/99264), [@palnabarun](https://github.com/palnabarun)) [SIG Auth, Instrumentation and Node] -- Migrate pkg/kubelet/util to structured logging ([#99823](https://github.com/kubernetes/kubernetes/pull/99823), [@navidshaikh](https://github.com/navidshaikh)) [SIG Instrumentation and Node] -- Migrate proxy/userspace/proxier.go logs to structured logging ([#97837](https://github.com/kubernetes/kubernetes/pull/97837), [@JornShen](https://github.com/JornShen)) [SIG Network] -- Migrate some kubelet/metrics log messages to structured logging ([#98627](https://github.com/kubernetes/kubernetes/pull/98627), [@jialaijun](https://github.com/jialaijun)) [SIG Instrumentation and Node] -- Process start time on Windows now uses current process information ([#97491](https://github.com/kubernetes/kubernetes/pull/97491), [@jsturtevant](https://github.com/jsturtevant)) [SIG API Machinery, CLI, Cloud Provider, Cluster Lifecycle, Instrumentation and Windows] - -### Uncategorized - -- Migrate pkg/kubelet/stats to structured logging ([#99607](https://github.com/kubernetes/kubernetes/pull/99607), [@krzysiekg](https://github.com/krzysiekg)) [SIG Node] -- The DownwardAPIHugePages feature is beta. Users may use the feature if all workers in their cluster are min 1.20 version. The feature will be enabled by default in all installations in 1.22. ([#99610](https://github.com/kubernetes/kubernetes/pull/99610), [@derekwaynecarr](https://github.com/derekwaynecarr)) [SIG Node] - -## Dependencies - -### Added -- github.com/go-errors/errors: [v1.0.1](https://github.com/go-errors/errors/tree/v1.0.1) -- github.com/gobuffalo/here: [v0.6.0](https://github.com/gobuffalo/here/tree/v0.6.0) -- github.com/google/shlex: [e7afc7f](https://github.com/google/shlex/tree/e7afc7f) -- github.com/markbates/pkger: [v0.17.1](https://github.com/markbates/pkger/tree/v0.17.1) -- github.com/monochromegane/go-gitignore: [205db1a](https://github.com/monochromegane/go-gitignore/tree/205db1a) -- github.com/niemeyer/pretty: [a10e7ca](https://github.com/niemeyer/pretty/tree/a10e7ca) -- github.com/xlab/treeprint: [a009c39](https://github.com/xlab/treeprint/tree/a009c39) -- go.starlark.net: 8dd3e2e -- golang.org/x/term: 6a3ed07 -- sigs.k8s.io/kustomize/api: v0.8.5 -- sigs.k8s.io/kustomize/cmd/config: v0.9.7 -- sigs.k8s.io/kustomize/kustomize/v4: v4.0.5 -- sigs.k8s.io/kustomize/kyaml: v0.10.15 - -### Changed -- dmitri.shuralyov.com/gpu/mtl: 666a987 → 28db891 -- github.com/creack/pty: [v1.1.7 → v1.1.9](https://github.com/creack/pty/compare/v1.1.7...v1.1.9) -- github.com/go-openapi/spec: [v0.19.3 → v0.19.5](https://github.com/go-openapi/spec/compare/v0.19.3...v0.19.5) -- github.com/go-openapi/strfmt: [v0.19.3 → v0.19.5](https://github.com/go-openapi/strfmt/compare/v0.19.3...v0.19.5) -- github.com/go-openapi/validate: [v0.19.5 → v0.19.8](https://github.com/go-openapi/validate/compare/v0.19.5...v0.19.8) -- github.com/google/cadvisor: [v0.38.7 → v0.38.8](https://github.com/google/cadvisor/compare/v0.38.7...v0.38.8) -- github.com/kr/text: [v0.1.0 → v0.2.0](https://github.com/kr/text/compare/v0.1.0...v0.2.0) -- github.com/mattn/go-runewidth: [v0.0.2 → v0.0.7](https://github.com/mattn/go-runewidth/compare/v0.0.2...v0.0.7) -- github.com/olekukonko/tablewriter: [a0225b3 → v0.0.4](https://github.com/olekukonko/tablewriter/compare/a0225b3...v0.0.4) -- github.com/sergi/go-diff: [v1.0.0 → v1.1.0](https://github.com/sergi/go-diff/compare/v1.0.0...v1.1.0) -- golang.org/x/crypto: 7f63de1 → 5ea612d -- golang.org/x/exp: 6cc2880 → 85be41e -- golang.org/x/mobile: d2bd2a2 → e6ae53a -- golang.org/x/mod: v0.3.0 → ce943fd -- golang.org/x/net: 69a7880 → 3d97a24 -- golang.org/x/sys: 5cba982 → a50acf3 -- golang.org/x/time: 3af7569 → f8bda1e -- golang.org/x/tools: 113979e → v0.1.0 -- gopkg.in/check.v1: 41f04d3 → 8fa4692 -- gopkg.in/yaml.v2: v2.2.8 → v2.4.0 -- k8s.io/kube-openapi: d219536 → 591a79e -- k8s.io/system-validators: v1.3.0 → v1.4.0 - -### Removed -- github.com/codegangsta/negroni: [v1.0.0](https://github.com/codegangsta/negroni/tree/v1.0.0) -- github.com/golangplus/bytes: [45c989f](https://github.com/golangplus/bytes/tree/45c989f) -- github.com/golangplus/fmt: [2a5d6d7](https://github.com/golangplus/fmt/tree/2a5d6d7) -- github.com/gorilla/context: [v1.1.1](https://github.com/gorilla/context/tree/v1.1.1) -- github.com/kr/pty: [v1.1.5](https://github.com/kr/pty/tree/v1.1.5) -- sigs.k8s.io/kustomize: v2.0.3+incompatible - - - -# v1.21.0-beta.0 - - -## Downloads for v1.21.0-beta.0 - -### Source Code - -filename | sha512 hash --------- | ----------- -[kubernetes.tar.gz](https://dl.k8s.io/v1.21.0-beta.0/kubernetes.tar.gz) | 69b73a03b70b0ed006e9fef3f5b9bc68f0eb8dc40db6cc04777c03a2cb83a008c783012ca186b1c48357fb192403dbcf6960f120924785e2076e215b9012d546 -[kubernetes-src.tar.gz](https://dl.k8s.io/v1.21.0-beta.0/kubernetes-src.tar.gz) | 9620fb6d37634271bdd423c09f33f3bd29e74298aa82c47dffc8cb6bd2ff44fa8987a53c53bc529db4ca96ec41503aa81cc8d0c3ac106f3b06c4720de933a8e6 - -### Client binaries - -filename | sha512 hash --------- | ----------- -[kubernetes-client-darwin-amd64.tar.gz](https://dl.k8s.io/v1.21.0-beta.0/kubernetes-client-darwin-amd64.tar.gz) | 2a6f3fcd6b571f5ccde56b91e6e179a01899244be496dae16a2a16e0405c9437b75c6dc853b56f9a4876a7c0a60ec624ccd28400bf8fb960258263172f6860ba -[kubernetes-client-linux-386.tar.gz](https://dl.k8s.io/v1.21.0-beta.0/kubernetes-client-linux-386.tar.gz) | 78fe9ad9f9a9bc043293327223f0038a2c087ca65e87187a6dcae7a24aef9565fe498d295a4639b0b90524469a04930022fcecd815d0afc742eb87ddd8eb7ef5 -[kubernetes-client-linux-amd64.tar.gz](https://dl.k8s.io/v1.21.0-beta.0/kubernetes-client-linux-amd64.tar.gz) | c025f5e5bd132355e7dd1296cf2ec752264e7f754c4d95fc34b076bd75bef2f571d30872bcb3d138ce95c592111353d275a80eb31f82c07000874b4c56282dbd -[kubernetes-client-linux-arm.tar.gz](https://dl.k8s.io/v1.21.0-beta.0/kubernetes-client-linux-arm.tar.gz) | 9975cd2f08fbc202575fb15ba6fc51dab23155ca4d294ebb48516a81efa51f58bab3a87d41c865103756189b554c020371d729ad42880ba788f25047ffc46910 -[kubernetes-client-linux-arm64.tar.gz](https://dl.k8s.io/v1.21.0-beta.0/kubernetes-client-linux-arm64.tar.gz) | 56a6836e24471e42e9d9a8488453f2d55598d70c8aca0a307d5116139c930c25c469fd0d1ab5060fbe88dad75a9b5209a08dc11d644af5f3ebebfbcb6c16266c -[kubernetes-client-linux-ppc64le.tar.gz](https://dl.k8s.io/v1.21.0-beta.0/kubernetes-client-linux-ppc64le.tar.gz) | b6a6cc9baad0ad85ed079ee80e6d6acc905095cfb440998bbc0f553b94fa80077bd58b8692754de477517663d51161705e6e89a1b6d04aa74819800db3517722 -[kubernetes-client-linux-s390x.tar.gz](https://dl.k8s.io/v1.21.0-beta.0/kubernetes-client-linux-s390x.tar.gz) | 7b743481b340f510bf9ae28ea8ea91150aa1e8c37fe104b66d7b3aff62f5e6db3c590d2c13d14dbb5c928de31c7613372def2496075853611d10d6b5fa5b60bd -[kubernetes-client-windows-386.tar.gz](https://dl.k8s.io/v1.21.0-beta.0/kubernetes-client-windows-386.tar.gz) | df06c7a524ce84c1f8d7836aa960c550c88dbca0ec4854df4dd0a85b3c84b8ecbc41b54e8c4669ce28ac670659ff0fad795deb1bc539f3c3b3aa885381265f5a -[kubernetes-client-windows-amd64.tar.gz](https://dl.k8s.io/v1.21.0-beta.0/kubernetes-client-windows-amd64.tar.gz) | 4568497b684564f2a94fbea6cbfd778b891231470d9a6956c3b7a3268643d13b855c0fc5ebea5f769300cc0c7719c2c331c387f468816f182f63e515adeaa7a0 - -### Server binaries - -filename | sha512 hash --------- | ----------- -[kubernetes-server-linux-amd64.tar.gz](https://dl.k8s.io/v1.21.0-beta.0/kubernetes-server-linux-amd64.tar.gz) | 42883cca2d312153baf693fc6024a295359a421e74fd70eefc927413be4e0353debe634e7cca6b9a8f7d8a0cee3717e03ba5d29a306e93139b1c2f3027535a6d -[kubernetes-server-linux-arm.tar.gz](https://dl.k8s.io/v1.21.0-beta.0/kubernetes-server-linux-arm.tar.gz) | e0042215e84c769ba4fc4d159ccf67b2c4a26206bfffb0ec5152723dc813ff9c1426aa0e9b963d7bfa2efb266ca43561b596b459152882ebb42102ccf60bd8eb -[kubernetes-server-linux-arm64.tar.gz](https://dl.k8s.io/v1.21.0-beta.0/kubernetes-server-linux-arm64.tar.gz) | bfad29d43e14152cb9bc7c4df6aa77929c6eca64a294bb832215bdba9fa0ee2195a2b709c0267dc7426bb371b547ee80bb8461a8c678c9bffa0819aa7db96289 -[kubernetes-server-linux-ppc64le.tar.gz](https://dl.k8s.io/v1.21.0-beta.0/kubernetes-server-linux-ppc64le.tar.gz) | ca67674c01c6cebdc8160c85b449eab1a23bb0557418665246e0208543fa2eaaf97679685c7b49bee3a4300904c0399c3d762ae34dc3e279fd69ce792c4b07ff -[kubernetes-server-linux-s390x.tar.gz](https://dl.k8s.io/v1.21.0-beta.0/kubernetes-server-linux-s390x.tar.gz) | 285352b628ec754b01b8ad4ef1427223a142d58ebcb46f6861df14d68643133b32330460b213b1ba5bc5362ff2b6dacd8e0c2d20cce6e760fa1954af8a60df8b - -### Node binaries - -filename | sha512 hash --------- | ----------- -[kubernetes-node-linux-amd64.tar.gz](https://dl.k8s.io/v1.21.0-beta.0/kubernetes-node-linux-amd64.tar.gz) | d92d9b30e7e44134a0cd9db4c01924d365991ea16b3131200b02a82cff89c8701f618cd90e7f1c65427bd4bb5f78b10d540b2262de2c143b401fa44e5b25627b -[kubernetes-node-linux-arm.tar.gz](https://dl.k8s.io/v1.21.0-beta.0/kubernetes-node-linux-arm.tar.gz) | 551092f23c27fdea4bb2d0547f6075892534892a96fc2be7786f82b58c93bffdb5e1c20f8f11beb8bed46c24f36d4c18ec5ac9755435489efa28e6ae775739bd -[kubernetes-node-linux-arm64.tar.gz](https://dl.k8s.io/v1.21.0-beta.0/kubernetes-node-linux-arm64.tar.gz) | 26ae7f4163e527349b8818ee38b9ee062314ab417f307afa49c146df8f5a2bd689509b128bd4a1efd3896fd89571149a9955ada91f8ca0c2f599cd863d613c86 -[kubernetes-node-linux-ppc64le.tar.gz](https://dl.k8s.io/v1.21.0-beta.0/kubernetes-node-linux-ppc64le.tar.gz) | 821fa953f6cebc69d2d481e489f3e90899813d20e2eefbabbcadd019d004108e7540f741fabe60e8e7c6adbb1053ac97898bbdddec3ca19f34a71aa3312e0d4e -[kubernetes-node-linux-s390x.tar.gz](https://dl.k8s.io/v1.21.0-beta.0/kubernetes-node-linux-s390x.tar.gz) | 22197d4f66205d5aa9de83dfddcc4f2bb3195fd7067cdb5c21e61dbeae217bc112fb7ecff8a539579b60ad92298c2b4c87b9b7c7e6ec1ee1ffa0c6e4bc4412c1 -[kubernetes-node-windows-amd64.tar.gz](https://dl.k8s.io/v1.21.0-beta.0/kubernetes-node-windows-amd64.tar.gz) | 7e22e0d9603562a04dee16a513579f06b1ff6354d97d669bd68f8777ec7f89f6ef027fb23ab0445d7bba0bb689352f0cc748ce90e3f597c6ebe495464a96b860 - -## Changelog since v1.21.0-alpha.3 - -## Urgent Upgrade Notes - -### (No, really, you MUST read this before you upgrade) - - - The metric `storage_operation_errors_total` is not removed, but is marked deprecated, and the metric `storage_operation_status_count` is marked deprecated. In both cases the storage_operation_duration_seconds metric can be used to recover equivalent counts (using `status=fail-unknown` in the case of `storage_operations_errors_total`). ([#99045](https://github.com/kubernetes/kubernetes/pull/99045), [@mattcary](https://github.com/mattcary)) [SIG Instrumentation and Storage] - -## Changes by Kind - -### Deprecation - -- The `batch/v2alpha1` CronJob type definitions and clients are deprecated and removed. ([#96987](https://github.com/kubernetes/kubernetes/pull/96987), [@soltysh](https://github.com/soltysh)) [SIG API Machinery, Apps, CLI and Testing] - -### API Change - -- Cluster admins can now turn off /debug/pprof and /debug/flags/v endpoint in kubelet by setting enableProfilingHandler and enableDebugFlagsHandler to false in their kubelet configuration file. enableProfilingHandler and enableDebugFlagsHandler can be set to true only when enableDebuggingHandlers is also set to true. ([#98458](https://github.com/kubernetes/kubernetes/pull/98458), [@SaranBalaji90](https://github.com/SaranBalaji90)) [SIG Node] -- The BoundServiceAccountTokenVolume feature has been promoted to beta, and enabled by default. - - This changes the tokens provided to containers at `/var/run/secrets/kubernetes.io/serviceaccount/token` to be time-limited, auto-refreshed, and invalidated when the containing pod is deleted. - - Clients should reload the token from disk periodically (once per minute is recommended) to ensure they continue to use a valid token. `k8s.io/client-go` version v11.0.0+ and v0.15.0+ reload tokens automatically. - - By default, injected tokens are given an extended lifetime so they remain valid even after a new refreshed token is provided. The metric `serviceaccount_stale_tokens_total` can be used to monitor for workloads that are depending on the extended lifetime and are continuing to use tokens even after a refreshed token is provided to the container. If that metric indicates no existing workloads are depending on extended lifetimes, injected token lifetime can be shortened to 1 hour by starting `kube-apiserver` with `--service-account-extend-token-expiration=false`. ([#95667](https://github.com/kubernetes/kubernetes/pull/95667), [@zshihang](https://github.com/zshihang)) [SIG API Machinery, Auth, Cluster Lifecycle and Testing] - -### Feature - -- A new histogram metric to track the time it took to delete a job by the ttl-after-finished controller ([#98676](https://github.com/kubernetes/kubernetes/pull/98676), [@ahg-g](https://github.com/ahg-g)) [SIG Apps and Instrumentation] -- AWS cloudprovider supports auto-discovering subnets without any kubernetes.io/cluster/ tags. It also supports additional service annotation service.beta.kubernetes.io/aws-load-balancer-subnets to manually configure the subnets. ([#97431](https://github.com/kubernetes/kubernetes/pull/97431), [@kishorj](https://github.com/kishorj)) [SIG Cloud Provider] -- Add --permit-address-sharing flag to kube-apiserver to listen with SO_REUSEADDR. While allowing to listen on wildcard IPs like 0.0.0.0 and specific IPs in parallel, it avoid waiting for the kernel to release socket in TIME_WAIT state, and hence, considably reducing kube-apiserver restart times under certain conditions. ([#93861](https://github.com/kubernetes/kubernetes/pull/93861), [@sttts](https://github.com/sttts)) [SIG API Machinery] -- Add `csi_operations_seconds` metric on kubelet that exposes CSI operations duration and status for node CSI operations. ([#98979](https://github.com/kubernetes/kubernetes/pull/98979), [@Jiawei0227](https://github.com/Jiawei0227)) [SIG Instrumentation and Storage] -- Add `migrated` field into `storage_operation_duration_seconds` metric ([#99050](https://github.com/kubernetes/kubernetes/pull/99050), [@Jiawei0227](https://github.com/Jiawei0227)) [SIG Apps, Instrumentation and Storage] -- Add bash-completion for comma separated list on `kubectl get` ([#98301](https://github.com/kubernetes/kubernetes/pull/98301), [@phil9909](https://github.com/phil9909)) [SIG CLI] -- Added support for installing arm64 node artifacts. ([#99242](https://github.com/kubernetes/kubernetes/pull/99242), [@liu-cong](https://github.com/liu-cong)) [SIG Cloud Provider] -- Feature gate RootCAConfigMap is graduated to GA in 1.21 and will be removed in 1.22. ([#98033](https://github.com/kubernetes/kubernetes/pull/98033), [@zshihang](https://github.com/zshihang)) [SIG API Machinery and Auth] -- Kubeadm: during "init" and "join" perform preflight validation on the host / node name and throw warnings if a name is not compliant ([#99194](https://github.com/kubernetes/kubernetes/pull/99194), [@pacoxu](https://github.com/pacoxu)) [SIG Cluster Lifecycle] -- Kubectl: `kubectl get` will omit managed fields by default now. Users could set `--show-managed-fields` to true to show managedFields when the output format is either `json` or `yaml`. ([#96878](https://github.com/kubernetes/kubernetes/pull/96878), [@knight42](https://github.com/knight42)) [SIG CLI and Testing] -- Metrics can now be disabled explicitly via a command line flag (i.e. '--disabled-metrics=bad_metric1,bad_metric2') ([#99217](https://github.com/kubernetes/kubernetes/pull/99217), [@logicalhan](https://github.com/logicalhan)) [SIG API Machinery, Cluster Lifecycle and Instrumentation] -- TTLAfterFinished is now beta and enabled by default ([#98678](https://github.com/kubernetes/kubernetes/pull/98678), [@ahg-g](https://github.com/ahg-g)) [SIG Apps and Auth] -- The `RunAsGroup` feature has been promoted to GA in this release. ([#94641](https://github.com/kubernetes/kubernetes/pull/94641), [@krmayankk](https://github.com/krmayankk)) [SIG Auth and Node] -- Turn CronJobControllerV2 on by default. ([#98878](https://github.com/kubernetes/kubernetes/pull/98878), [@soltysh](https://github.com/soltysh)) [SIG Apps] -- UDP protocol support for Agnhost connect subcommand ([#98639](https://github.com/kubernetes/kubernetes/pull/98639), [@knabben](https://github.com/knabben)) [SIG Testing] -- Upgrades `IPv6Dualstack` to `Beta` and turns it on by default. Clusters new and existing will not be affected until user starting adding secondary pod and service cidrs cli flags as described here: https://github.com/kubernetes/enhancements/tree/master/keps/sig-network/563-dual-stack ([#98969](https://github.com/kubernetes/kubernetes/pull/98969), [@khenidak](https://github.com/khenidak)) [SIG API Machinery, Apps, Cloud Provider, Network and Node] - -### Documentation - -- Fix ALPHA stability level reference link ([#98641](https://github.com/kubernetes/kubernetes/pull/98641), [@Jeffwan](https://github.com/Jeffwan)) [SIG Auth, Cloud Provider, Instrumentation and Storage] - -### Failing Test - -- Escape the special characters like `[`, `]` and ` ` that exist in vsphere windows path ([#98830](https://github.com/kubernetes/kubernetes/pull/98830), [@liyanhui1228](https://github.com/liyanhui1228)) [SIG Storage and Windows] -- Kube-proxy: fix a bug on UDP NodePort Services where stale conntrack entries may blackhole the traffic directed to the NodePort. ([#98305](https://github.com/kubernetes/kubernetes/pull/98305), [@aojea](https://github.com/aojea)) [SIG Network] - -### Bug or Regression - -- Add missing --kube-api-content-type in kubemark hollow template ([#98911](https://github.com/kubernetes/kubernetes/pull/98911), [@Jeffwan](https://github.com/Jeffwan)) [SIG Scalability and Testing] -- Avoid duplicate error messages when runing kubectl edit quota ([#98201](https://github.com/kubernetes/kubernetes/pull/98201), [@pacoxu](https://github.com/pacoxu)) [SIG API Machinery and Apps] -- Cleanup subnet in frontend IP configs to prevent huge subnet request bodies in some scenarios. ([#98133](https://github.com/kubernetes/kubernetes/pull/98133), [@nilo19](https://github.com/nilo19)) [SIG Cloud Provider] -- Fix errors when accessing Windows container stats for Dockershim ([#98510](https://github.com/kubernetes/kubernetes/pull/98510), [@jsturtevant](https://github.com/jsturtevant)) [SIG Node and Windows] -- Fixes spurious errors about IPv6 in kube-proxy logs on nodes with IPv6 disabled. ([#99127](https://github.com/kubernetes/kubernetes/pull/99127), [@danwinship](https://github.com/danwinship)) [SIG Network and Node] -- In the method that ensures that the docker and containerd are in the correct containers with the proper OOM score set up, fixed the bug of identifying containerd process. ([#97888](https://github.com/kubernetes/kubernetes/pull/97888), [@pacoxu](https://github.com/pacoxu)) [SIG Node] -- Kubelet now cleans up orphaned volume directories automatically ([#95301](https://github.com/kubernetes/kubernetes/pull/95301), [@lorenz](https://github.com/lorenz)) [SIG Node and Storage] -- When dynamically provisioning Azure File volumes for a premium account, the requested size will be set to 100GB if the request is initially lower than this value to accommodate Azure File requirements. ([#99122](https://github.com/kubernetes/kubernetes/pull/99122), [@huffmanca](https://github.com/huffmanca)) [SIG Cloud Provider and Storage] - -### Other (Cleanup or Flake) - -- APIs for kubelet annotations and labels from k8s.io/kubernetes/pkg/kubelet/apis are now available under k8s.io/kubelet/pkg/apis/ ([#98931](https://github.com/kubernetes/kubernetes/pull/98931), [@michaelbeaumont](https://github.com/michaelbeaumont)) [SIG Apps, Auth and Node] -- Migrate `pkg/kubelet/(pod, pleg)` to structured logging ([#98990](https://github.com/kubernetes/kubernetes/pull/98990), [@gjkim42](https://github.com/gjkim42)) [SIG Instrumentation and Node] -- Migrate pkg/kubelet/nodestatus to structured logging ([#99001](https://github.com/kubernetes/kubernetes/pull/99001), [@QiWang19](https://github.com/QiWang19)) [SIG Node] -- Migrate pkg/kubelet/server logs to structured logging ([#98643](https://github.com/kubernetes/kubernetes/pull/98643), [@chenyw1990](https://github.com/chenyw1990)) [SIG Node] -- Migrate proxy/winkernel/proxier.go logs to structured logging ([#98001](https://github.com/kubernetes/kubernetes/pull/98001), [@JornShen](https://github.com/JornShen)) [SIG Network and Windows] -- Migrate scheduling_queue.go to structured logging ([#98358](https://github.com/kubernetes/kubernetes/pull/98358), [@tanjing2020](https://github.com/tanjing2020)) [SIG Scheduling] -- Several flags related to the deprecated dockershim which are present in the kubelet command line are now deprecated. ([#98730](https://github.com/kubernetes/kubernetes/pull/98730), [@dims](https://github.com/dims)) [SIG Node] -- The deprecated feature gates `CSIDriverRegistry`, `BlockVolume` and `CSIBlockVolume` are now unconditionally enabled and can no longer be specified in component invocations. ([#98021](https://github.com/kubernetes/kubernetes/pull/98021), [@gavinfish](https://github.com/gavinfish)) [SIG Storage] - -## Dependencies - -### Added -_Nothing has changed._ - -### Changed -- sigs.k8s.io/structured-merge-diff/v4: v4.0.2 → v4.0.3 - -### Removed -_Nothing has changed._ - - - -# v1.21.0-alpha.3 - - -## Downloads for v1.21.0-alpha.3 - -### Source Code - -filename | sha512 hash --------- | ----------- -[kubernetes.tar.gz](https://dl.k8s.io/v1.21.0-alpha.3/kubernetes.tar.gz) | 704ec916a1dbd134c54184d2652671f80ae09274f9d23dbbed312944ebeccbc173e2e6b6949b38bdbbfdaf8aa032844deead5efeda1b3150f9751386d9184bc8 -[kubernetes-src.tar.gz](https://dl.k8s.io/v1.21.0-alpha.3/kubernetes-src.tar.gz) | 57db9e7560cfc9c10e7059cb5faf9c4bd5eb8f9b7964f44f000a417021cf80873184b774e7c66c80d4aba84c14080c6bc335618db3d2e5f276436ae065e25408 - -### Client binaries - -filename | sha512 hash --------- | ----------- -[kubernetes-client-darwin-amd64.tar.gz](https://dl.k8s.io/v1.21.0-alpha.3/kubernetes-client-darwin-amd64.tar.gz) | e2706efda92d5cf4f8b69503bb2f7703a8754407eff7f199bb77847838070e720e5f572126c14daa4c0c03b59bb1a63c1dfdeb6e936a40eff1d5497e871e3409 -[kubernetes-client-linux-386.tar.gz](https://dl.k8s.io/v1.21.0-alpha.3/kubernetes-client-linux-386.tar.gz) | 007bb23c576356ed0890bdfd25a0f98d552599e0ffec19fb982591183c7c1f216d8a3ffa3abf15216be12ae5c4b91fdcd48a7306a2d26b007b86a6abd553fc61 -[kubernetes-client-linux-amd64.tar.gz](https://dl.k8s.io/v1.21.0-alpha.3/kubernetes-client-linux-amd64.tar.gz) | 39504b0c610348beba60e8866fff265bad58034f74504951cd894c151a248db718d10f77ebc83f2c38b2d517f8513a46325b38889eefa261ca6dbffeceba50ff -[kubernetes-client-linux-arm.tar.gz](https://dl.k8s.io/v1.21.0-alpha.3/kubernetes-client-linux-arm.tar.gz) | 30bc2c40d0c759365422ad1651a6fb35909be771f463c5b971caf401f9209525d05256ab70c807e88628dd357c2896745eecf13eda0b748464da97d0a5ef2066 -[kubernetes-client-linux-arm64.tar.gz](https://dl.k8s.io/v1.21.0-alpha.3/kubernetes-client-linux-arm64.tar.gz) | 085cdf574dc8fd33ece667130b8c45830b522a07860e03a2384283b1adea73a9652ef3dfaa566e69ee00aea1a6461608814b3ce7a3f703e4a934304f7ae12f97 -[kubernetes-client-linux-ppc64le.tar.gz](https://dl.k8s.io/v1.21.0-alpha.3/kubernetes-client-linux-ppc64le.tar.gz) | b34b845037d83ea7b3e2d80a9ede4f889b71b17b93b1445f0d936a36e98c13ed6ada125630a68d9243a5fcd311ee37cdcc0c05da484da8488ea5060bc529dbfc -[kubernetes-client-linux-s390x.tar.gz](https://dl.k8s.io/v1.21.0-alpha.3/kubernetes-client-linux-s390x.tar.gz) | c4758adc7a404b776556efaa79655db2a70777c562145d6ea6887f3335988367a0c2fcd4383e469340f2a768b22e786951de212805ca1cb91104d41c21e0c9ce -[kubernetes-client-windows-386.tar.gz](https://dl.k8s.io/v1.21.0-alpha.3/kubernetes-client-windows-386.tar.gz) | f51edc79702bbd1d9cb3a672852a405e11b20feeab64c5411a7e85c9af304960663eb6b23ef96e0f8c44a722fecf58cb6d700ea2c42c05b3269d8efd5ad803f2 -[kubernetes-client-windows-amd64.tar.gz](https://dl.k8s.io/v1.21.0-alpha.3/kubernetes-client-windows-amd64.tar.gz) | 6a3507ce4ac40a0dc7e4720538863fa15f8faf025085a032f34b8fa0f6fa4e8c26849baf649b5b32829b9182e04f82721b13950d31cf218c35be6bf1c05d6abf - -### Server binaries - -filename | sha512 hash --------- | ----------- -[kubernetes-server-linux-amd64.tar.gz](https://dl.k8s.io/v1.21.0-alpha.3/kubernetes-server-linux-amd64.tar.gz) | 19181d162dfb0b30236e2bf1111000e037eece87c037ca2b24622ca94cb88db86aa4da4ca533522518b209bc9983bbfd6b880a7898e0da96b33f3f6c4690539b -[kubernetes-server-linux-arm.tar.gz](https://dl.k8s.io/v1.21.0-alpha.3/kubernetes-server-linux-arm.tar.gz) | 42a02f9e08a78ad5da6e5fa1ab12bf1e3c967c472fdbdadbd8746586da74dc8093682ba8513ff2a5301393c47ee9021b860e88ada56b13da386ef485708e46ca -[kubernetes-server-linux-arm64.tar.gz](https://dl.k8s.io/v1.21.0-alpha.3/kubernetes-server-linux-arm64.tar.gz) | 3c8ba8eb02f70061689bd7fab7813542005efe2edc6cfc6b7aecd03ffedf0b81819ad91d69fff588e83023d595eefbfe636aa55e1856add8733bf42fff3c748f -[kubernetes-server-linux-ppc64le.tar.gz](https://dl.k8s.io/v1.21.0-alpha.3/kubernetes-server-linux-ppc64le.tar.gz) | cd9e6537450411c39a06fd0b5819db3d16b668d403fb3627ec32c0e32dd1c4860e942934578ca0e1d1b8e6f21f450ff81e37e0cd46ff5c5faf7847ab074aefc5 -[kubernetes-server-linux-s390x.tar.gz](https://dl.k8s.io/v1.21.0-alpha.3/kubernetes-server-linux-s390x.tar.gz) | ada3f65e53bc0e0c0229694dd48c425388089d6d77111a62476d1b08f6ad1d8ab3d60b9ed7d95ac1b42c2c6be8dc0618f40679717160769743c43583d8452362 - -### Node binaries - -filename | sha512 hash --------- | ----------- -[kubernetes-node-linux-amd64.tar.gz](https://dl.k8s.io/v1.21.0-alpha.3/kubernetes-node-linux-amd64.tar.gz) | ae0fec6aa59e49624b55d9a11c12fdf717ddfe04bdfd4f69965d03004a34e52ee4a3e83f7b61d0c6a86f43b72c99f3decb195b39ae529ef30526d18ec5f58f83 -[kubernetes-node-linux-arm.tar.gz](https://dl.k8s.io/v1.21.0-alpha.3/kubernetes-node-linux-arm.tar.gz) | 9a48c140ab53b7ed8ecec6903988a1a474efc16d2538e5974bc9a12f0c9190be78c4f9e326bf4e982d0b7045a80b99dd0fda7e9b650663be5b89bfd991596746 -[kubernetes-node-linux-arm64.tar.gz](https://dl.k8s.io/v1.21.0-alpha.3/kubernetes-node-linux-arm64.tar.gz) | 6912adbc9300344bea470d6435f7b387bfce59767078c11728ce59faf47cd3f72b41b9604fcc5cda45e9816fe939fbe2fb33e52a773e6ff2dfa9a615b4df6141 -[kubernetes-node-linux-ppc64le.tar.gz](https://dl.k8s.io/v1.21.0-alpha.3/kubernetes-node-linux-ppc64le.tar.gz) | d66dccfe3e6ed6d81567c70703f15375a53992b3a5e2814b98c32e581b861ad95912e03ed2562415d087624c008038bb4a816611fa255442ae752968ea15856b -[kubernetes-node-linux-s390x.tar.gz](https://dl.k8s.io/v1.21.0-alpha.3/kubernetes-node-linux-s390x.tar.gz) | ad8c69a28f1fbafa3f1cb54909bfd3fc22b104bed63d7ca2b296208c9d43eb5f2943a0ff267da4c185186cdd9f7f77b315cd7f5f1bf9858c0bf42eceb9ac3c58 -[kubernetes-node-windows-amd64.tar.gz](https://dl.k8s.io/v1.21.0-alpha.3/kubernetes-node-windows-amd64.tar.gz) | 91d723aa848a9cb028f5bcb41090ca346fb973961521d025c4399164de2c8029b57ca2c4daca560d3c782c05265d2eb0edb0abcce6f23d3efbecf2316a54d650 - -## Changelog since v1.21.0-alpha.2 - -## Urgent Upgrade Notes - -### (No, really, you MUST read this before you upgrade) - - - Newly provisioned PVs by gce-pd will no longer have the beta FailureDomain label. gce-pd volume plugin will start to have GA topology label instead. ([#98700](https://github.com/kubernetes/kubernetes/pull/98700), [@Jiawei0227](https://github.com/Jiawei0227)) [SIG Cloud Provider, Storage and Testing] - - Remove alpha CSIMigrationXXComplete flag and add alpha InTreePluginXXUnregister flag. Deprecate CSIMigrationvSphereComplete flag and it will be removed in 1.22. ([#98243](https://github.com/kubernetes/kubernetes/pull/98243), [@Jiawei0227](https://github.com/Jiawei0227)) [SIG Node and Storage] - -## Changes by Kind - -### API Change - -- Adds support for portRange / EndPort in Network Policy ([#97058](https://github.com/kubernetes/kubernetes/pull/97058), [@rikatz](https://github.com/rikatz)) [SIG Apps and Network] -- Fixes using server-side apply with APIService resources ([#98576](https://github.com/kubernetes/kubernetes/pull/98576), [@kevindelgado](https://github.com/kevindelgado)) [SIG API Machinery, Apps and Testing] -- Kubernetes is now built using go1.15.7 ([#98363](https://github.com/kubernetes/kubernetes/pull/98363), [@cpanato](https://github.com/cpanato)) [SIG Cloud Provider, Instrumentation, Node, Release and Testing] -- Scheduler extender filter interface now can report unresolvable failed nodes in the new field `FailedAndUnresolvableNodes` of `ExtenderFilterResult` struct. Nodes in this map will be skipped in the preemption phase. ([#92866](https://github.com/kubernetes/kubernetes/pull/92866), [@cofyc](https://github.com/cofyc)) [SIG Scheduling] - -### Feature - -- A lease can only attach up to 10k objects. ([#98257](https://github.com/kubernetes/kubernetes/pull/98257), [@lingsamuel](https://github.com/lingsamuel)) [SIG API Machinery] -- Add ignore-errors flag for drain, support none-break drain in group ([#98203](https://github.com/kubernetes/kubernetes/pull/98203), [@yuzhiquan](https://github.com/yuzhiquan)) [SIG CLI] -- Base-images: Update to debian-iptables:buster-v1.4.0 - - Uses iptables 1.8.5 - - base-images: Update to debian-base:buster-v1.3.0 - - cluster/images/etcd: Build etcd:3.4.13-2 image - - Uses debian-base:buster-v1.3.0 ([#98401](https://github.com/kubernetes/kubernetes/pull/98401), [@pacoxu](https://github.com/pacoxu)) [SIG Testing] -- Export NewDebuggingRoundTripper function and DebugLevel options in the k8s.io/client-go/transport package. ([#98324](https://github.com/kubernetes/kubernetes/pull/98324), [@atosatto](https://github.com/atosatto)) [SIG API Machinery] -- Kubectl wait ensures that observedGeneration >= generation if applicable ([#97408](https://github.com/kubernetes/kubernetes/pull/97408), [@KnicKnic](https://github.com/KnicKnic)) [SIG CLI] -- Kubernetes is now built using go1.15.8 ([#98834](https://github.com/kubernetes/kubernetes/pull/98834), [@cpanato](https://github.com/cpanato)) [SIG Cloud Provider, Instrumentation, Release and Testing] -- New admission controller "denyserviceexternalips" is available. Clusters which do not *need- the Service "externalIPs" feature should enable this controller and be more secure. ([#97395](https://github.com/kubernetes/kubernetes/pull/97395), [@thockin](https://github.com/thockin)) [SIG API Machinery] -- Overall, enable the feature of `PreferNominatedNode` will improve the performance of scheduling where preemption might frequently happen, but in theory, enable the feature of `PreferNominatedNode`, the pod might not be scheduled to the best candidate node in the cluster. ([#93179](https://github.com/kubernetes/kubernetes/pull/93179), [@chendave](https://github.com/chendave)) [SIG Scheduling and Testing] -- Pause image upgraded to 3.4.1 in kubelet and kubeadm for both Linux and Windows. ([#98205](https://github.com/kubernetes/kubernetes/pull/98205), [@pacoxu](https://github.com/pacoxu)) [SIG CLI, Cloud Provider, Cluster Lifecycle, Node, Testing and Windows] -- The `ServiceAccountIssuerDiscovery` feature has graduated to GA, and is unconditionally enabled. The `ServiceAccountIssuerDiscovery` feature-gate will be removed in 1.22. ([#98553](https://github.com/kubernetes/kubernetes/pull/98553), [@mtaufen](https://github.com/mtaufen)) [SIG API Machinery, Auth and Testing] - -### Documentation - -- Feat: azure file migration go beta in 1.21. Feature gates CSIMigration to Beta (on by default) and CSIMigrationAzureFile to Beta (off by default since it requires installation of the AzureFile CSI Driver) - The in-tree AzureFile plugin "kubernetes.io/azure-file" is now deprecated and will be removed in 1.23. Users should enable CSIMigration + CSIMigrationAzureFile features and install the AzureFile CSI Driver (https://github.com/kubernetes-sigs/azurefile-csi-driver) to avoid disruption to existing Pod and PVC objects at that time. - Users should start using the AzureFile CSI Driver directly for any new volumes. ([#96293](https://github.com/kubernetes/kubernetes/pull/96293), [@andyzhangx](https://github.com/andyzhangx)) [SIG Cloud Provider] - -### Failing Test - -- Kubelet: the HostPort implementation in dockershim was not taking into consideration the HostIP field, causing that the same HostPort can not be used with different IP addresses. - This bug causes the conformance test "HostPort validates that there is no conflict between pods with same hostPort but different hostIP and protocol" to fail. ([#98755](https://github.com/kubernetes/kubernetes/pull/98755), [@aojea](https://github.com/aojea)) [SIG Cloud Provider, Network and Node] - -### Bug or Regression - -- Fix NPE in ephemeral storage eviction ([#98261](https://github.com/kubernetes/kubernetes/pull/98261), [@wzshiming](https://github.com/wzshiming)) [SIG Node] -- Fixed a bug that on k8s nodes, when the policy of INPUT chain in filter table is not ACCEPT, healthcheck nodeport would not work. - Added iptables rules to allow healthcheck nodeport traffic. ([#97824](https://github.com/kubernetes/kubernetes/pull/97824), [@hanlins](https://github.com/hanlins)) [SIG Network] -- Fixed kube-proxy container image architecture for non amd64 images. ([#98526](https://github.com/kubernetes/kubernetes/pull/98526), [@saschagrunert](https://github.com/saschagrunert)) [SIG API Machinery, Release and Testing] -- Fixed provisioning of Cinder volumes migrated to CSI when StorageClass with AllowedTopologies was used. ([#98311](https://github.com/kubernetes/kubernetes/pull/98311), [@jsafrane](https://github.com/jsafrane)) [SIG Storage] -- Fixes a panic in the disruption budget controller for PDB objects with invalid selectors ([#98750](https://github.com/kubernetes/kubernetes/pull/98750), [@mortent](https://github.com/mortent)) [SIG Apps] -- Fixes connection errors when using `--volume-host-cidr-denylist` or `--volume-host-allow-local-loopback` ([#98436](https://github.com/kubernetes/kubernetes/pull/98436), [@liggitt](https://github.com/liggitt)) [SIG Network and Storage] -- If the user specifies an invalid timeout in the request URL, the request will be aborted with an HTTP 400. - - in cases where the client specifies a timeout in the request URL, the overall request deadline is shortened now since the deadline is setup as soon as the request is received by the apiserver. ([#96901](https://github.com/kubernetes/kubernetes/pull/96901), [@tkashem](https://github.com/tkashem)) [SIG API Machinery and Testing] -- Kubeadm: Some text in the `kubeadm upgrade plan` output has changed. If you have scripts or other automation that parses this output, please review these changes and update your scripts to account for the new output. ([#98728](https://github.com/kubernetes/kubernetes/pull/98728), [@stmcginnis](https://github.com/stmcginnis)) [SIG Cluster Lifecycle] -- Kubeadm: fix a bug where external credentials in an existing admin.conf prevented the CA certificate to be written in the cluster-info ConfigMap. ([#98882](https://github.com/kubernetes/kubernetes/pull/98882), [@kvaps](https://github.com/kvaps)) [SIG Cluster Lifecycle] -- Kubeadm: fix bad token placeholder text in "config print *-defaults --help" ([#98839](https://github.com/kubernetes/kubernetes/pull/98839), [@Mattias-](https://github.com/Mattias-)) [SIG Cluster Lifecycle] -- Kubeadm: get k8s CI version markers from k8s infra bucket ([#98836](https://github.com/kubernetes/kubernetes/pull/98836), [@hasheddan](https://github.com/hasheddan)) [SIG Cluster Lifecycle and Release] -- Mitigate CVE-2020-8555 for kube-up using GCE by preventing local loopback folume hosts. ([#97934](https://github.com/kubernetes/kubernetes/pull/97934), [@mattcary](https://github.com/mattcary)) [SIG Cloud Provider and Storage] -- Remove CSI topology from migrated in-tree gcepd volume. ([#97823](https://github.com/kubernetes/kubernetes/pull/97823), [@Jiawei0227](https://github.com/Jiawei0227)) [SIG Cloud Provider and Storage] -- Sync node status during kubelet node shutdown. - Adds an pod admission handler that rejects new pods when the node is in progress of shutting down. ([#98005](https://github.com/kubernetes/kubernetes/pull/98005), [@wzshiming](https://github.com/wzshiming)) [SIG Node] -- Truncates a message if it hits the NoteLengthLimit when the scheduler records an event for the pod that indicates the pod has failed to schedule. ([#98715](https://github.com/kubernetes/kubernetes/pull/98715), [@carlory](https://github.com/carlory)) [SIG Scheduling] -- We will no longer automatically delete all data when a failure is detected during creation of the volume data file on a CSI volume. Now we will only remove the data file and volume path. ([#96021](https://github.com/kubernetes/kubernetes/pull/96021), [@huffmanca](https://github.com/huffmanca)) [SIG Storage] - -### Other (Cleanup or Flake) - -- Fix the description of command line flags that can override --config ([#98254](https://github.com/kubernetes/kubernetes/pull/98254), [@changshuchao](https://github.com/changshuchao)) [SIG Scheduling] -- Migrate scheduler/taint_manager.go structured logging ([#98259](https://github.com/kubernetes/kubernetes/pull/98259), [@tanjing2020](https://github.com/tanjing2020)) [SIG Apps] -- Migrate staging/src/k8s.io/apiserver/pkg/admission logs to structured logging ([#98138](https://github.com/kubernetes/kubernetes/pull/98138), [@lala123912](https://github.com/lala123912)) [SIG API Machinery] -- Resolves flakes in the Ingress conformance tests due to conflicts with controllers updating the Ingress object ([#98430](https://github.com/kubernetes/kubernetes/pull/98430), [@liggitt](https://github.com/liggitt)) [SIG Network and Testing] -- The default delegating authorization options now allow unauthenticated access to healthz, readyz, and livez. A system:masters user connecting to an authz delegator will not perform an authz check. ([#98325](https://github.com/kubernetes/kubernetes/pull/98325), [@deads2k](https://github.com/deads2k)) [SIG API Machinery, Auth, Cloud Provider and Scheduling] -- The e2e suite can be instructed not to wait for pods in kube-system to be ready or for all nodes to be ready by passing `--allowed-not-ready-nodes=-1` when invoking the e2e.test program. This allows callers to run subsets of the e2e suite in scenarios other than perfectly healthy clusters. ([#98781](https://github.com/kubernetes/kubernetes/pull/98781), [@smarterclayton](https://github.com/smarterclayton)) [SIG Testing] -- The feature gates `WindowsGMSA` and `WindowsRunAsUserName` that are GA since v1.18 are now removed. ([#96531](https://github.com/kubernetes/kubernetes/pull/96531), [@ialidzhikov](https://github.com/ialidzhikov)) [SIG Node and Windows] -- The new `-gce-zones` flag on the `e2e.test` binary instructs tests that check for information about how the cluster interacts with the cloud to limit their queries to the provided zone list. If not specified, the current behavior of asking the cloud provider for all available zones in multi zone clusters is preserved. ([#98787](https://github.com/kubernetes/kubernetes/pull/98787), [@smarterclayton](https://github.com/smarterclayton)) [SIG API Machinery, Cluster Lifecycle and Testing] - -## Dependencies - -### Added -- github.com/moby/spdystream: [v0.2.0](https://github.com/moby/spdystream/tree/v0.2.0) - -### Changed -- github.com/NYTimes/gziphandler: [56545f4 → v1.1.1](https://github.com/NYTimes/gziphandler/compare/56545f4...v1.1.1) -- github.com/container-storage-interface/spec: [v1.2.0 → v1.3.0](https://github.com/container-storage-interface/spec/compare/v1.2.0...v1.3.0) -- github.com/go-logr/logr: [v0.2.0 → v0.4.0](https://github.com/go-logr/logr/compare/v0.2.0...v0.4.0) -- github.com/gogo/protobuf: [v1.3.1 → v1.3.2](https://github.com/gogo/protobuf/compare/v1.3.1...v1.3.2) -- github.com/kisielk/errcheck: [v1.2.0 → v1.5.0](https://github.com/kisielk/errcheck/compare/v1.2.0...v1.5.0) -- github.com/yuin/goldmark: [v1.1.27 → v1.2.1](https://github.com/yuin/goldmark/compare/v1.1.27...v1.2.1) -- golang.org/x/sync: cd5d95a → 67f06af -- golang.org/x/tools: c1934b7 → 113979e -- k8s.io/klog/v2: v2.4.0 → v2.5.0 -- sigs.k8s.io/apiserver-network-proxy/konnectivity-client: v0.0.14 → v0.0.15 - -### Removed -- github.com/docker/spdystream: [449fdfc](https://github.com/docker/spdystream/tree/449fdfc) - - - -# v1.21.0-alpha.2 - - -## Downloads for v1.21.0-alpha.2 - -### Source Code - -filename | sha512 hash --------- | ----------- -[kubernetes.tar.gz](https://dl.k8s.io/v1.21.0-alpha.2/kubernetes.tar.gz) | 6836f6c8514253fe0831fd171fc4ed92eb6d9a773491c8dc82b90d171a1b10076bd6bfaea56ec1e199c5f46c273265bdb9f174f0b2d99c5af1de4c99b862329e -[kubernetes-src.tar.gz](https://dl.k8s.io/v1.21.0-alpha.2/kubernetes-src.tar.gz) | d137694804741a05ab09e5f9a418448b66aba0146c028eafce61bcd9d7c276521e345ce9223ffbc703e8172041d58dfc56a3242a4df3686f24905a4541fcd306 - -### Client binaries - -filename | sha512 hash --------- | ----------- -[kubernetes-client-darwin-amd64.tar.gz](https://dl.k8s.io/v1.21.0-alpha.2/kubernetes-client-darwin-amd64.tar.gz) | 9478b047a97717953f365c13a098feb7e3cb30a3df22e1b82aa945f2208dcc5cb90afc441ba059a3ae7aafb4ee000ec3a52dc65a8c043a5ac7255a391c875330 -[kubernetes-client-linux-386.tar.gz](https://dl.k8s.io/v1.21.0-alpha.2/kubernetes-client-linux-386.tar.gz) | 44c8dd4b1ddfc256d35786c8abf45b0eb5f0794f5e310d2efc865748adddc50e8bf38aa71295ae8a82884cb65f2e0b9b0737b000f96fd8f2d5c19971d7c4d8e8 -[kubernetes-client-linux-amd64.tar.gz](https://dl.k8s.io/v1.21.0-alpha.2/kubernetes-client-linux-amd64.tar.gz) | e1291989892769de6b978c17b8612b94da6f3b735a4d895100af622ca9ebb968c75548afea7ab00445869625dd0da3afec979e333afbb445805f5d31c1c13cc7 -[kubernetes-client-linux-arm.tar.gz](https://dl.k8s.io/v1.21.0-alpha.2/kubernetes-client-linux-arm.tar.gz) | 3c4bcb8cbe73822d68a2f62553a364e20bec56b638c71d0f58679b4f4b277d809142346f18506914e694f6122a3e0f767eab20b7b1c4dbb79e4c5089981ae0f1 -[kubernetes-client-linux-arm64.tar.gz](https://dl.k8s.io/v1.21.0-alpha.2/kubernetes-client-linux-arm64.tar.gz) | 9389974a790268522e187f5ba5237f3ee4684118c7db76bc3d4164de71d8208702747ec333b204c7a78073ab42553cbbce13a1883fab4fec617e093b05fab332 -[kubernetes-client-linux-ppc64le.tar.gz](https://dl.k8s.io/v1.21.0-alpha.2/kubernetes-client-linux-ppc64le.tar.gz) | 63399e53a083b5af3816c28ff162c9de6b64c75da4647f0d6bbaf97afdf896823cb1e556f2abac75c6516072293026d3ff9f30676fd75143ac6ca3f4d21f4327 -[kubernetes-client-linux-s390x.tar.gz](https://dl.k8s.io/v1.21.0-alpha.2/kubernetes-client-linux-s390x.tar.gz) | 50898f197a9d923971ff9046c9f02779b57f7b3cea7da02f3ea9bab8c08d65a9c4a7531a2470fa14783460f52111a52b96ebf916c0a1d8215b4070e4e861c1b0 -[kubernetes-client-windows-386.tar.gz](https://dl.k8s.io/v1.21.0-alpha.2/kubernetes-client-windows-386.tar.gz) | a7743e839e1aa19f5ee20b6ee5000ac8ef9e624ac5be63bb574fad6992e4b9167193ed07e03c9bc524e88bfeed66c95341a38a03bff1b10bc9910345f33019f0 -[kubernetes-client-windows-amd64.tar.gz](https://dl.k8s.io/v1.21.0-alpha.2/kubernetes-client-windows-amd64.tar.gz) | 5f1d19c230bd3542866d16051808d184e9dd3e2f8c001ed4cee7b5df91f872380c2bf56a3add8c9413ead9d8c369efce2bcab4412174df9b823d3592677bf74e - -### Server binaries - -filename | sha512 hash --------- | ----------- -[kubernetes-server-linux-amd64.tar.gz](https://dl.k8s.io/v1.21.0-alpha.2/kubernetes-server-linux-amd64.tar.gz) | ef2cac10febde231aeb6f131e589450c560eeaab8046b49504127a091cddc17bc518c2ad56894a6a033033ab6fc6e121b1cc23691683bc36f45fe6b1dd8e0510 -[kubernetes-server-linux-arm.tar.gz](https://dl.k8s.io/v1.21.0-alpha.2/kubernetes-server-linux-arm.tar.gz) | d11c9730307f08e80b2b8a7c64c3e9a9e43c622002e377dfe3a386f4541e24adc79a199a6f280f40298bb36793194fd44ed45defe8a3ee54a9cb1386bc26e905 -[kubernetes-server-linux-arm64.tar.gz](https://dl.k8s.io/v1.21.0-alpha.2/kubernetes-server-linux-arm64.tar.gz) | 28f8c32bf98ee1add7edf5d341c3bac1afc0085f90dcbbfb8b27a92087f13e2b53c327c8935ee29bf1dc3160655b32bbe3e29d5741a8124a3848a777e7d42933 -[kubernetes-server-linux-ppc64le.tar.gz](https://dl.k8s.io/v1.21.0-alpha.2/kubernetes-server-linux-ppc64le.tar.gz) | 99ae8d44b0de3518c27fa8bbddd2ecf053dfb789fb9d65f8a4ecf4c8331cf63d2f09a41c2bcd5573247d5f66a1b2e51944379df1715017d920d521b98589508a -[kubernetes-server-linux-s390x.tar.gz](https://dl.k8s.io/v1.21.0-alpha.2/kubernetes-server-linux-s390x.tar.gz) | f8c0e954a2dfc6845614488dadeed069cc7f3f08e33c351d7a77c6ef97867af590932e8576d12998a820a0e4d35d2eee797c764e2810f09ab1e90a5acaeaad33 - -### Node binaries - -filename | sha512 hash --------- | ----------- -[kubernetes-node-linux-amd64.tar.gz](https://dl.k8s.io/v1.21.0-alpha.2/kubernetes-node-linux-amd64.tar.gz) | c5456d50bfbe0d75fb150b3662ed7468a0abd3970792c447824f326894382c47bbd3a2cc5a290f691c8c09585ff6fe505ab86b4aff2b7e5ccee11b5e6354ae6c -[kubernetes-node-linux-arm.tar.gz](https://dl.k8s.io/v1.21.0-alpha.2/kubernetes-node-linux-arm.tar.gz) | 335b5cd8672e053302fd94d932fb2fa2e48eeeb1799650b3f93acdfa635e03a8453637569ab710c46885c8317759f4c60aaaf24dca9817d9fa47500fe4a3ca53 -[kubernetes-node-linux-arm64.tar.gz](https://dl.k8s.io/v1.21.0-alpha.2/kubernetes-node-linux-arm64.tar.gz) | 3ee87dbeed8ace9351ac89bdaf7274dd10b4faec3ceba0825f690ec7a2bb7eb7c634274a1065a0939eec8ff3e43f72385f058f4ec141841550109e775bc5eff9 -[kubernetes-node-linux-ppc64le.tar.gz](https://dl.k8s.io/v1.21.0-alpha.2/kubernetes-node-linux-ppc64le.tar.gz) | 6956f965b8d719b164214ec9195fdb2c776b907fe6d2c524082f00c27872a73475927fd7d2a994045ce78f6ad2aa5aeaf1eb5514df1810d2cfe342fd4e5ce4a1 -[kubernetes-node-linux-s390x.tar.gz](https://dl.k8s.io/v1.21.0-alpha.2/kubernetes-node-linux-s390x.tar.gz) | 3b643aa905c709c57083c28dd9e8ffd88cb64466cda1499da7fc54176b775003e08b9c7a07b0964064df67c8142f6f1e6c13bfc261bd65fb064049920bfa57d0 -[kubernetes-node-windows-amd64.tar.gz](https://dl.k8s.io/v1.21.0-alpha.2/kubernetes-node-windows-amd64.tar.gz) | b2e6d6fb0091f2541f9925018c2bdbb0138a95bab06b4c6b38abf4b7144b2575422263b78fb3c6fd09e76d90a25a8d35a6d4720dc169794d42c95aa22ecc6d5f - -## Changelog since v1.21.0-alpha.1 - -## Urgent Upgrade Notes - -### (No, really, you MUST read this before you upgrade) - - - Remove storage metrics `storage_operation_errors_total`, since we already have `storage_operation_status_count`.And add new field `status` for `storage_operation_duration_seconds`, so that we can know about all status storage operation latency. ([#98332](https://github.com/kubernetes/kubernetes/pull/98332), [@JornShen](https://github.com/JornShen)) [SIG Instrumentation and Storage] - -## Changes by Kind - -### Deprecation - -- Remove the TokenRequest and TokenRequestProjection feature gates ([#97148](https://github.com/kubernetes/kubernetes/pull/97148), [@wawa0210](https://github.com/wawa0210)) [SIG Node] -- Removing experimental windows container hyper-v support with Docker ([#97141](https://github.com/kubernetes/kubernetes/pull/97141), [@wawa0210](https://github.com/wawa0210)) [SIG Node and Windows] -- The `export` query parameter (inconsistently supported by API resources and deprecated in v1.14) is fully removed. Requests setting this query parameter will now receive a 400 status response. ([#98312](https://github.com/kubernetes/kubernetes/pull/98312), [@deads2k](https://github.com/deads2k)) [SIG API Machinery, Auth and Testing] - -### API Change - -- Enable SPDY pings to keep connections alive, so that `kubectl exec` and `kubectl port-forward` won't be interrupted. ([#97083](https://github.com/kubernetes/kubernetes/pull/97083), [@knight42](https://github.com/knight42)) [SIG API Machinery and CLI] - -### Documentation - -- Official support to build kubernetes with docker-machine / remote docker is removed. This change does not affect building kubernetes with docker locally. ([#97935](https://github.com/kubernetes/kubernetes/pull/97935), [@adeniyistephen](https://github.com/adeniyistephen)) [SIG Release and Testing] -- Set kubelet option `--volume-stats-agg-period` to negative value to disable volume calculations. ([#96675](https://github.com/kubernetes/kubernetes/pull/96675), [@pacoxu](https://github.com/pacoxu)) [SIG Node] - -### Bug or Regression - -- Clean ReplicaSet by revision instead of creation timestamp in deployment controller ([#97407](https://github.com/kubernetes/kubernetes/pull/97407), [@waynepeking348](https://github.com/waynepeking348)) [SIG Apps] -- Ensure that client-go's EventBroadcaster is safe (non-racy) during shutdown. ([#95664](https://github.com/kubernetes/kubernetes/pull/95664), [@DirectXMan12](https://github.com/DirectXMan12)) [SIG API Machinery] -- Fix azure file migration issue ([#97877](https://github.com/kubernetes/kubernetes/pull/97877), [@andyzhangx](https://github.com/andyzhangx)) [SIG Auth, Cloud Provider and Storage] -- Fix kubelet from panic after getting the wrong signal ([#98200](https://github.com/kubernetes/kubernetes/pull/98200), [@wzshiming](https://github.com/wzshiming)) [SIG Node] -- Fix repeatedly acquire the inhibit lock ([#98088](https://github.com/kubernetes/kubernetes/pull/98088), [@wzshiming](https://github.com/wzshiming)) [SIG Node] -- Fixed a bug that the kubelet cannot start on BtrfS. ([#98042](https://github.com/kubernetes/kubernetes/pull/98042), [@gjkim42](https://github.com/gjkim42)) [SIG Node] -- Fixed an issue with garbage collection failing to clean up namespaced children of an object also referenced incorrectly by cluster-scoped children ([#98068](https://github.com/kubernetes/kubernetes/pull/98068), [@liggitt](https://github.com/liggitt)) [SIG API Machinery and Apps] -- Fixed no effect namespace when exposing deployment with --dry-run=client. ([#97492](https://github.com/kubernetes/kubernetes/pull/97492), [@masap](https://github.com/masap)) [SIG CLI] -- Fixing a bug where a failed node may not have the NoExecute taint set correctly ([#96876](https://github.com/kubernetes/kubernetes/pull/96876), [@howieyuen](https://github.com/howieyuen)) [SIG Apps and Node] -- Indentation of `Resource Quota` block in kubectl describe namespaces output gets correct. ([#97946](https://github.com/kubernetes/kubernetes/pull/97946), [@dty1er](https://github.com/dty1er)) [SIG CLI] -- KUBECTL_EXTERNAL_DIFF now accepts equal sign for additional parameters. ([#98158](https://github.com/kubernetes/kubernetes/pull/98158), [@dougsland](https://github.com/dougsland)) [SIG CLI] -- Kubeadm: fix a bug where "kubeadm join" would not properly handle missing names for existing etcd members. ([#97372](https://github.com/kubernetes/kubernetes/pull/97372), [@ihgann](https://github.com/ihgann)) [SIG Cluster Lifecycle] -- Kubelet should ignore cgroup driver check on Windows node. ([#97764](https://github.com/kubernetes/kubernetes/pull/97764), [@pacoxu](https://github.com/pacoxu)) [SIG Node and Windows] -- Make podTopologyHints protected by lock ([#95111](https://github.com/kubernetes/kubernetes/pull/95111), [@choury](https://github.com/choury)) [SIG Node] -- Readjust kubelet_containers_per_pod_count bucket ([#98169](https://github.com/kubernetes/kubernetes/pull/98169), [@wawa0210](https://github.com/wawa0210)) [SIG Instrumentation and Node] -- Scores from InterPodAffinity have stronger differentiation. ([#98096](https://github.com/kubernetes/kubernetes/pull/98096), [@leileiwan](https://github.com/leileiwan)) [SIG Scheduling] -- Specifying the KUBE_TEST_REPO environment variable when e2e tests are executed will instruct the test infrastructure to load that image from a location within the specified repo, using a predefined pattern. ([#93510](https://github.com/kubernetes/kubernetes/pull/93510), [@smarterclayton](https://github.com/smarterclayton)) [SIG Testing] -- Static pods will be deleted gracefully. ([#98103](https://github.com/kubernetes/kubernetes/pull/98103), [@gjkim42](https://github.com/gjkim42)) [SIG Node] -- Use network.Interface.VirtualMachine.ID to get the binded VM - Skip standalone VM when reconciling LoadBalancer ([#97635](https://github.com/kubernetes/kubernetes/pull/97635), [@nilo19](https://github.com/nilo19)) [SIG Cloud Provider] - -### Other (Cleanup or Flake) - -- Kubeadm: change the default image repository for CI images from 'gcr.io/kubernetes-ci-images' to 'gcr.io/k8s-staging-ci-images' ([#97087](https://github.com/kubernetes/kubernetes/pull/97087), [@SataQiu](https://github.com/SataQiu)) [SIG Cluster Lifecycle] -- Migrate generic_scheduler.go and types.go to structured logging. ([#98134](https://github.com/kubernetes/kubernetes/pull/98134), [@tanjing2020](https://github.com/tanjing2020)) [SIG Scheduling] -- Migrate proxy/winuserspace/proxier.go logs to structured logging ([#97941](https://github.com/kubernetes/kubernetes/pull/97941), [@JornShen](https://github.com/JornShen)) [SIG Network] -- Migrate staging/src/k8s.io/apiserver/pkg/audit/policy/reader.go logs to structured logging. ([#98252](https://github.com/kubernetes/kubernetes/pull/98252), [@lala123912](https://github.com/lala123912)) [SIG API Machinery and Auth] -- Migrate staging\src\k8s.io\apiserver\pkg\endpoints logs to structured logging ([#98093](https://github.com/kubernetes/kubernetes/pull/98093), [@lala123912](https://github.com/lala123912)) [SIG API Machinery] -- Node ([#96552](https://github.com/kubernetes/kubernetes/pull/96552), [@pandaamanda](https://github.com/pandaamanda)) [SIG Apps, Cloud Provider, Node and Scheduling] -- The kubectl alpha debug command was scheduled to be removed in v1.21. ([#98111](https://github.com/kubernetes/kubernetes/pull/98111), [@pandaamanda](https://github.com/pandaamanda)) [SIG CLI] -- Update cri-tools to [v1.20.0](https://github.com/kubernetes-sigs/cri-tools/releases/tag/v1.20.0) ([#97967](https://github.com/kubernetes/kubernetes/pull/97967), [@rajibmitra](https://github.com/rajibmitra)) [SIG Cloud Provider] -- Windows nodes on GCE will take longer to start due to dependencies installed at node creation time. ([#98284](https://github.com/kubernetes/kubernetes/pull/98284), [@pjh](https://github.com/pjh)) [SIG Cloud Provider] - -## Dependencies - -### Added -_Nothing has changed._ - -### Changed -- github.com/google/cadvisor: [v0.38.6 → v0.38.7](https://github.com/google/cadvisor/compare/v0.38.6...v0.38.7) -- k8s.io/gengo: 83324d8 → b6c5ce2 - -### Removed -_Nothing has changed._ - - - -# v1.21.0-alpha.1 - - -## Downloads for v1.21.0-alpha.1 - -### Source Code - -filename | sha512 hash --------- | ----------- -[kubernetes.tar.gz](https://dl.k8s.io/v1.21.0-alpha.1/kubernetes.tar.gz) | b2bacd5c3fc9f829e6269b7d2006b0c6e464ff848bb0a2a8f2fe52ad2d7c4438f099bd8be847d8d49ac6e4087f4d74d5c3a967acd798e0b0cb4d7a2bdb122997 -[kubernetes-src.tar.gz](https://dl.k8s.io/v1.21.0-alpha.1/kubernetes-src.tar.gz) | 518ac5acbcf23902fb1b902b69dbf3e86deca5d8a9b5f57488a15f185176d5a109558f3e4df062366af874eca1bcd61751ee8098b0beb9bcdc025d9a1c9be693 - -### Client binaries - -filename | sha512 hash --------- | ----------- -[kubernetes-client-darwin-amd64.tar.gz](https://dl.k8s.io/v1.21.0-alpha.1/kubernetes-client-darwin-amd64.tar.gz) | eaa7aea84a5ed954df5ec710cbeb6ec88b46465f43cb3d09aabe2f714b84a050a50bf5736089f09dbf1090f2e19b44823d656c917e3c8c877630756c3026f2b6 -[kubernetes-client-linux-386.tar.gz](https://dl.k8s.io/v1.21.0-alpha.1/kubernetes-client-linux-386.tar.gz) | 47f74b8d46ad1779c5b0b5f15aa15d5513a504eeb6f53db4201fbe9ff8956cb986b7c1b0e9d50a99f78e9e2a7f304f3fc1cc2fa239296d9a0dd408eb6069e975 -[kubernetes-client-linux-amd64.tar.gz](https://dl.k8s.io/v1.21.0-alpha.1/kubernetes-client-linux-amd64.tar.gz) | 1a148e282628b008c8abd03dd12ec177ced17584b5115d92cd33dd251e607097d42e9da8c7089bd947134b900f85eb75a4740b6a5dd580c105455b843559df39 -[kubernetes-client-linux-arm.tar.gz](https://dl.k8s.io/v1.21.0-alpha.1/kubernetes-client-linux-arm.tar.gz) | d13d2feb73bd032dc01f7e2955b98d8215a39fe1107d037a73fa1f7d06c3b93ebaa53ed4952d845c64454ef3cca533edb97132d234d50b6fb3bcbd8a8ad990eb -[kubernetes-client-linux-arm64.tar.gz](https://dl.k8s.io/v1.21.0-alpha.1/kubernetes-client-linux-arm64.tar.gz) | 8252105a17b09a78e9ad2c024e4e401a69764ac869708a071aaa06f81714c17b9e7c5b2eb8efde33f24d0b59f75c5da607d5e1e72bdf12adfbb8c829205cd1c1 -[kubernetes-client-linux-ppc64le.tar.gz](https://dl.k8s.io/v1.21.0-alpha.1/kubernetes-client-linux-ppc64le.tar.gz) | 297a9082df4988389dc4be30eb636dff49f36f5d87047bab44745884e610f46a17ae3a08401e2cab155b7c439f38057bfd8288418215f7dd3bf6a49dbe61ea0e -[kubernetes-client-linux-s390x.tar.gz](https://dl.k8s.io/v1.21.0-alpha.1/kubernetes-client-linux-s390x.tar.gz) | 04c06490dd17cd5dccfd92bafa14acf64280ceaea370d9635f23aeb6984d1beae6d0d1d1506edc6f30f927deeb149b989d3e482b47fbe74008b371f629656e79 -[kubernetes-client-windows-386.tar.gz](https://dl.k8s.io/v1.21.0-alpha.1/kubernetes-client-windows-386.tar.gz) | ec6e9e87a7d685f8751d7e58f24f417753cff5554a7229218cb3a08195d461b2e12409344950228e9fbbc92a8a06d35dd86242da6ff1e6652ec1fae0365a88c1 -[kubernetes-client-windows-amd64.tar.gz](https://dl.k8s.io/v1.21.0-alpha.1/kubernetes-client-windows-amd64.tar.gz) | 51039e6221d3126b5d15e797002ae01d4f0b10789c5d2056532f27ef13f35c5a2e51be27764fda68e8303219963126559023aed9421313bec275c0827fbcaf8a - -### Server binaries - -filename | sha512 hash --------- | ----------- -[kubernetes-server-linux-amd64.tar.gz](https://dl.k8s.io/v1.21.0-alpha.1/kubernetes-server-linux-amd64.tar.gz) | 4edf820930c88716263560275e3bd7fadb8dc3700b9f8e1d266562e356e0abeb1a913f536377dab91218e3940b447d6bf1da343b85da25c2256dc4dcde5798dd -[kubernetes-server-linux-arm.tar.gz](https://dl.k8s.io/v1.21.0-alpha.1/kubernetes-server-linux-arm.tar.gz) | b15213e53a8ab4ba512ce6ef9ad42dd197d419c61615cd23de344227fd846c90448d8f3d98e555b63ba5b565afa627cca6b7e3990ebbbba359c96f2391302df1 -[kubernetes-server-linux-arm64.tar.gz](https://dl.k8s.io/v1.21.0-alpha.1/kubernetes-server-linux-arm64.tar.gz) | 5be29cca9a9358fc68351ee63e99d57dc2ffce6e42fc3345753dbbf7542ff2d770c4852424158540435fa6e097ce3afa9b13affc40c8b3b69fe8406798f8068f -[kubernetes-server-linux-ppc64le.tar.gz](https://dl.k8s.io/v1.21.0-alpha.1/kubernetes-server-linux-ppc64le.tar.gz) | 89fd99ab9ce85db0b94b86709932105efc883cc93959cf7ea9a39e79a4acea23064d7010eeb577450cccabe521c04b7ba47bbec212ed37edeed7cb04bad34518 -[kubernetes-server-linux-s390x.tar.gz](https://dl.k8s.io/v1.21.0-alpha.1/kubernetes-server-linux-s390x.tar.gz) | 2fbc30862c77d247aa8d96ab9d1a144599505287b0033a3a2d0988958e7bb2f2e8b67f52c1fec74b4ec47d74ba22cd0f6cb5c4228acbaa72b1678d5fece0254d - -### Node binaries - -filename | sha512 hash --------- | ----------- -[kubernetes-node-linux-amd64.tar.gz](https://dl.k8s.io/v1.21.0-alpha.1/kubernetes-node-linux-amd64.tar.gz) | 95658d321a0a371c0900b401d1469d96915310afbc4e4b9b11f031438bb188513b57d5a60b5316c3b0c18f541cda6f0ac42f59a76495f8abc743a067115da23a -[kubernetes-node-linux-arm.tar.gz](https://dl.k8s.io/v1.21.0-alpha.1/kubernetes-node-linux-arm.tar.gz) | f375acfb42aad6c65b833c270e7e3acfe9cd1d6b2441c33874e77faae263957f7acfe86f1b71f14298118595e4cc6952c7dea0c832f7f2e72428336f13034362 -[kubernetes-node-linux-arm64.tar.gz](https://dl.k8s.io/v1.21.0-alpha.1/kubernetes-node-linux-arm64.tar.gz) | 43b4baccd58d74e7f48d096ab92f2bbbcdf47e30e7a3d2b56c6cc9f90002cfd4fefaac894f69bd5f9f4dbdb09a4749a77eb76b1b97d91746bd96fe94457879ab -[kubernetes-node-linux-ppc64le.tar.gz](https://dl.k8s.io/v1.21.0-alpha.1/kubernetes-node-linux-ppc64le.tar.gz) | e7962b522c6c7c14b9ee4c1d254d8bdd9846b2b33b0443fc9c4a41be6c40e5e6981798b720f0148f36263d5cc45d5a2bb1dd2f9ab2838e3d002e45b9bddeb7bf -[kubernetes-node-linux-s390x.tar.gz](https://dl.k8s.io/v1.21.0-alpha.1/kubernetes-node-linux-s390x.tar.gz) | 49ebc97f01829e65f7de15be00b882513c44782eaadd1b1825a227e3bd3c73cc6aca8345af05b303d8c43aa2cb944a069755b2709effb8cc22eae621d25d4ba5 -[kubernetes-node-windows-amd64.tar.gz](https://dl.k8s.io/v1.21.0-alpha.1/kubernetes-node-windows-amd64.tar.gz) | 6e0fd7724b09e6befbcb53b33574e97f2db089f2eee4bbf391abb7f043103a5e6e32e3014c0531b88f9a3ca88887bbc68625752c44326f98dd53adb3a6d1bed8 - -## Changelog since v1.20.0 - -## Urgent Upgrade Notes - -### (No, really, you MUST read this before you upgrade) - - - Kube-proxy's IPVS proxy mode no longer sets the net.ipv4.conf.all.route_localnet sysctl parameter. Nodes upgrading will have net.ipv4.conf.all.route_localnet set to 1 but new nodes will inherit the system default (usually 0). If you relied on any behavior requiring net.ipv4.conf.all.route_localnet, you must set ensure it is enabled as kube-proxy will no longer set it automatically. This change helps to further mitigate CVE-2020-8558. ([#92938](https://github.com/kubernetes/kubernetes/pull/92938), [@lbernail](https://github.com/lbernail)) [SIG Network and Release] - -## Changes by Kind - -### Deprecation - -- Deprecate the `topologyKeys` field in Service. This capability will be replaced with upcoming work around Topology Aware Subsetting and Service Internal Traffic Policy. ([#96736](https://github.com/kubernetes/kubernetes/pull/96736), [@andrewsykim](https://github.com/andrewsykim)) [SIG Apps] -- Kubeadm: deprecated command "alpha selfhosting pivot" is removed now. ([#97627](https://github.com/kubernetes/kubernetes/pull/97627), [@knight42](https://github.com/knight42)) [SIG Cluster Lifecycle] -- Kubeadm: graduate the command `kubeadm alpha kubeconfig user` to `kubeadm kubeconfig user`. The `kubeadm alpha kubeconfig user` command is deprecated now. ([#97583](https://github.com/kubernetes/kubernetes/pull/97583), [@knight42](https://github.com/knight42)) [SIG Cluster Lifecycle] -- Kubeadm: the "kubeadm alpha certs" command is removed now, please use "kubeadm certs" instead. ([#97706](https://github.com/kubernetes/kubernetes/pull/97706), [@knight42](https://github.com/knight42)) [SIG Cluster Lifecycle] -- Remove the deprecated metrics "scheduling_algorithm_preemption_evaluation_seconds" and "binding_duration_seconds", suggest to use "scheduler_framework_extension_point_duration_seconds" instead. ([#96447](https://github.com/kubernetes/kubernetes/pull/96447), [@chendave](https://github.com/chendave)) [SIG Cluster Lifecycle, Instrumentation, Scheduling and Testing] -- The PodSecurityPolicy API is deprecated in 1.21, and will no longer be served starting in 1.25. ([#97171](https://github.com/kubernetes/kubernetes/pull/97171), [@deads2k](https://github.com/deads2k)) [SIG Auth and CLI] - -### API Change - -- Change the APIVersion proto name of BoundObjectRef from aPIVersion to apiVersion. ([#97379](https://github.com/kubernetes/kubernetes/pull/97379), [@kebe7jun](https://github.com/kebe7jun)) [SIG Auth] -- Promote Immutable Secrets/ConfigMaps feature to Stable. - This allows to set `Immutable` field in Secrets or ConfigMap object to mark their contents as immutable. ([#97615](https://github.com/kubernetes/kubernetes/pull/97615), [@wojtek-t](https://github.com/wojtek-t)) [SIG Apps, Architecture, Node and Testing] - -### Feature - -- Add flag --lease-max-object-size and metric etcd_lease_object_counts for kube-apiserver to config and observe max objects attached to a single etcd lease. ([#97480](https://github.com/kubernetes/kubernetes/pull/97480), [@lingsamuel](https://github.com/lingsamuel)) [SIG API Machinery, Instrumentation and Scalability] -- Add flag --lease-reuse-duration-seconds for kube-apiserver to config etcd lease reuse duration. ([#97009](https://github.com/kubernetes/kubernetes/pull/97009), [@lingsamuel](https://github.com/lingsamuel)) [SIG API Machinery and Scalability] -- Adds the ability to pass --strict-transport-security-directives to the kube-apiserver to set the HSTS header appropriately. Be sure you understand the consequences to browsers before setting this field. ([#96502](https://github.com/kubernetes/kubernetes/pull/96502), [@249043822](https://github.com/249043822)) [SIG Auth] -- Kubeadm now includes CoreDNS v1.8.0. ([#96429](https://github.com/kubernetes/kubernetes/pull/96429), [@rajansandeep](https://github.com/rajansandeep)) [SIG Cluster Lifecycle] -- Kubeadm: add support for certificate chain validation. When using kubeadm in external CA mode, this allows an intermediate CA to be used to sign the certificates. The intermediate CA certificate must be appended to each signed certificate for this to work correctly. ([#97266](https://github.com/kubernetes/kubernetes/pull/97266), [@robbiemcmichael](https://github.com/robbiemcmichael)) [SIG Cluster Lifecycle] -- Kubeadm: amend the node kernel validation to treat CGROUP_PIDS, FAIR_GROUP_SCHED as required and CFS_BANDWIDTH, CGROUP_HUGETLB as optional ([#96378](https://github.com/kubernetes/kubernetes/pull/96378), [@neolit123](https://github.com/neolit123)) [SIG Cluster Lifecycle and Node] -- The Kubernetes pause image manifest list now contains an image for Windows Server 20H2. ([#97322](https://github.com/kubernetes/kubernetes/pull/97322), [@claudiubelu](https://github.com/claudiubelu)) [SIG Windows] -- The apimachinery util/net function used to detect the bind address `ResolveBindAddress()` - takes into consideration global ip addresses on loopback interfaces when: - - the host has default routes - - there are no global IPs on those interfaces. - in order to support more complex network scenarios like BGP Unnumbered RFC 5549 ([#95790](https://github.com/kubernetes/kubernetes/pull/95790), [@aojea](https://github.com/aojea)) [SIG Network] - -### Bug or Regression - -- ## Changelog - - ### General - - Fix priority expander falling back to a random choice even though there is a higher priority option to choose - - Clone `kubernetes/kubernetes` in `update-vendor.sh` shallowly, instead of fetching all revisions - - Speed up binpacking by reducing the number of PreFilter calls (call once per pod instead of #pods*#nodes times) - - Speed up finding unneeded nodes by 5x+ in very large clusters by reducing the number of PreFilter calls - - Expose `--max-nodes-total` as a metric - - Errors in `IncreaseSize` changed from type `apiError` to `cloudProviderError` - - Make `build-in-docker` and `test-in-docker` work on Linux systems with SELinux enabled - - Fix an error where existing nodes were not considered as destinations while finding place for pods in scale-down simulations - - Remove redundant log lines and reduce severity around parsing kubeEnv - - Don't treat nodes created by virtual kubelet as nodes from non-autoscaled node groups - - Remove redundant logging around calculating node utilization - - Add configurable `--network` and `--rm` flags for docker in `Makefile` - - Subtract DaemonSet pods' requests from node allocatable in the denominator while computing node utilization - - Include taints by condition when determining if a node is unready/still starting - - Fix `update-vendor.sh` to work on OSX and zsh - - Add best-effort eviction for DaemonSet pods while scaling down non-empty nodes - - Add build support for ARM64 - - ### AliCloud - - Add missing daemonsets and replicasets to ALI example cluster role - - ### Apache CloudStack - - Add support for Apache CloudStack - - ### AWS - - Regenerate list of EC2 instances - - Fix pricing endpoint in AWS China Region - - ### Azure - - Add optional jitter on initial VMSS VM cache refresh, keep the refreshes spread over time - - Serve from cache for the whole period of ongoing throttling - - Fix unwanted VMSS VMs cache invalidations - - Enforce setting the number of retries if cloud provider backoff is enabled - - Don't update capacity if VMSS provisioning state is updating - - Support allocatable resources overrides via VMSS tags - - Add missing stable labels in template nodes - - Proactively set instance status to deleting on node deletions - - ### Cluster API - - Migrate interaction with the API from using internal types to using Unstructured - - Improve tests to work better with constrained resources - - Add support for node autodiscovery - - Add support for `--cloud-config` - - Update group identifier to use for Cluster API annotations - - ### Exoscale - - Add support for Exoscale - - ### GCE - - Decrease the number of GCE Read Requests made while deleting nodes - - Base pricing of custom instances on their instance family type - - Add pricing information for missing machine types - - Add pricing information for different GPU types - - Ignore the new `topology.gke.io/zone` label when comparing groups - - Add missing stable labels to template nodes - - ### HuaweiCloud - - Add auto scaling group support - - Implement node group by AS - - Implement getting desired instance number of node group - - Implement increasing node group size - - Implement TemplateNodeInfo - - Implement caching instances - - ### IONOS - - Add support for IONOS - - ### Kubemark - - Skip non-kubemark nodes while computing node infos for node groups. - - ### Magnum - - Add Magnum support in the Cluster Autoscaler helm chart - - ### Packet - - Allow empty nodepools - - Add support for multiple nodepools - - Add pricing support - - ## Image - Image: `k8s.gcr.io/autoscaling/cluster-autoscaler:v1.20.0` ([#97011](https://github.com/kubernetes/kubernetes/pull/97011), [@towca](https://github.com/towca)) [SIG Cloud Provider] -- AcceleratorStats will be available in the Summary API of kubelet when cri_stats_provider is used. ([#96873](https://github.com/kubernetes/kubernetes/pull/96873), [@ruiwen-zhao](https://github.com/ruiwen-zhao)) [SIG Node] -- Add limited lines to log when having tail option ([#93920](https://github.com/kubernetes/kubernetes/pull/93920), [@zhouya0](https://github.com/zhouya0)) [SIG Node] -- Avoid systemd-logind loading configuration warning ([#97950](https://github.com/kubernetes/kubernetes/pull/97950), [@wzshiming](https://github.com/wzshiming)) [SIG Node] -- Cloud-controller-manager: routes controller should not depend on --allocate-node-cidrs ([#97029](https://github.com/kubernetes/kubernetes/pull/97029), [@andrewsykim](https://github.com/andrewsykim)) [SIG Cloud Provider and Testing] -- Copy annotations with empty value when deployment rolls back ([#94858](https://github.com/kubernetes/kubernetes/pull/94858), [@waynepeking348](https://github.com/waynepeking348)) [SIG Apps] -- Detach volumes from vSphere nodes not tracked by attach-detach controller ([#96689](https://github.com/kubernetes/kubernetes/pull/96689), [@gnufied](https://github.com/gnufied)) [SIG Cloud Provider and Storage] -- Fix kubectl label error when local=true is set. ([#97440](https://github.com/kubernetes/kubernetes/pull/97440), [@pandaamanda](https://github.com/pandaamanda)) [SIG CLI] -- Fix Azure file share not deleted issue when the namespace is deleted ([#97417](https://github.com/kubernetes/kubernetes/pull/97417), [@andyzhangx](https://github.com/andyzhangx)) [SIG Cloud Provider and Storage] -- Fix CVE-2020-8555 for Gluster client connections. ([#97922](https://github.com/kubernetes/kubernetes/pull/97922), [@liggitt](https://github.com/liggitt)) [SIG Storage] -- Fix counting error in service/nodeport/loadbalancer quota check ([#97451](https://github.com/kubernetes/kubernetes/pull/97451), [@pacoxu](https://github.com/pacoxu)) [SIG API Machinery, Network and Testing] -- Fix kubectl-convert import known versions ([#97754](https://github.com/kubernetes/kubernetes/pull/97754), [@wzshiming](https://github.com/wzshiming)) [SIG CLI and Testing] -- Fix missing cadvisor machine metrics. ([#97006](https://github.com/kubernetes/kubernetes/pull/97006), [@lingsamuel](https://github.com/lingsamuel)) [SIG Node] -- Fix nil VMSS name when setting service to auto mode ([#97366](https://github.com/kubernetes/kubernetes/pull/97366), [@nilo19](https://github.com/nilo19)) [SIG Cloud Provider] -- Fix the panic when kubelet registers if a node object already exists with no Status.Capacity or Status.Allocatable ([#95269](https://github.com/kubernetes/kubernetes/pull/95269), [@SataQiu](https://github.com/SataQiu)) [SIG Node] -- Fix the regression with the slow pods termination. Before this fix pods may take an additional time to terminate - up to one minute. Reversing the change that ensured that CNI resources cleaned up when the pod is removed on API server. ([#97980](https://github.com/kubernetes/kubernetes/pull/97980), [@SergeyKanzhelev](https://github.com/SergeyKanzhelev)) [SIG Node] -- Fix to recover CSI volumes from certain dangling attachments ([#96617](https://github.com/kubernetes/kubernetes/pull/96617), [@yuga711](https://github.com/yuga711)) [SIG Apps and Storage] -- Fix: azure file latency issue for metadata-heavy workloads ([#97082](https://github.com/kubernetes/kubernetes/pull/97082), [@andyzhangx](https://github.com/andyzhangx)) [SIG Cloud Provider and Storage] -- Fixed Cinder volume IDs on OpenStack Train ([#96673](https://github.com/kubernetes/kubernetes/pull/96673), [@jsafrane](https://github.com/jsafrane)) [SIG Cloud Provider] -- Fixed FibreChannel volume plugin corrupting filesystems on detach of multipath volumes. ([#97013](https://github.com/kubernetes/kubernetes/pull/97013), [@jsafrane](https://github.com/jsafrane)) [SIG Storage] -- Fixed a bug in kubelet that will saturate CPU utilization after containerd got restarted. ([#97174](https://github.com/kubernetes/kubernetes/pull/97174), [@hanlins](https://github.com/hanlins)) [SIG Node] -- Fixed bug in CPUManager with race on container map access ([#97427](https://github.com/kubernetes/kubernetes/pull/97427), [@klueska](https://github.com/klueska)) [SIG Node] -- Fixed cleanup of block devices when /var/lib/kubelet is a symlink. ([#96889](https://github.com/kubernetes/kubernetes/pull/96889), [@jsafrane](https://github.com/jsafrane)) [SIG Storage] -- GCE Internal LoadBalancer sync loop will now release the ILB IP address upon sync failure. An error in ILB forwarding rule creation will no longer leak IP addresses. ([#97740](https://github.com/kubernetes/kubernetes/pull/97740), [@prameshj](https://github.com/prameshj)) [SIG Cloud Provider and Network] -- Ignore update pod with no new images in alwaysPullImages admission controller ([#96668](https://github.com/kubernetes/kubernetes/pull/96668), [@pacoxu](https://github.com/pacoxu)) [SIG Apps, Auth and Node] -- Kubeadm now installs version 3.4.13 of etcd when creating a cluster with v1.19 ([#97244](https://github.com/kubernetes/kubernetes/pull/97244), [@pacoxu](https://github.com/pacoxu)) [SIG Cluster Lifecycle] -- Kubeadm: avoid detection of the container runtime for commands that do not need it ([#97625](https://github.com/kubernetes/kubernetes/pull/97625), [@pacoxu](https://github.com/pacoxu)) [SIG Cluster Lifecycle] -- Kubeadm: fix a bug in the host memory detection code on 32bit Linux platforms ([#97403](https://github.com/kubernetes/kubernetes/pull/97403), [@abelbarrera15](https://github.com/abelbarrera15)) [SIG Cluster Lifecycle] -- Kubeadm: fix a bug where "kubeadm upgrade" commands can fail if CoreDNS v1.8.0 is installed. ([#97919](https://github.com/kubernetes/kubernetes/pull/97919), [@neolit123](https://github.com/neolit123)) [SIG Cluster Lifecycle] -- Performance regression [#97685](https://github.com/kubernetes/kubernetes/issues/97685) has been fixed. ([#97860](https://github.com/kubernetes/kubernetes/pull/97860), [@MikeSpreitzer](https://github.com/MikeSpreitzer)) [SIG API Machinery] -- Remove deprecated --cleanup-ipvs flag of kube-proxy, and make --cleanup flag always to flush IPVS ([#97336](https://github.com/kubernetes/kubernetes/pull/97336), [@maaoBit](https://github.com/maaoBit)) [SIG Network] -- The current version of the container image publicly exposed IP serving a /metrics endpoint to the Internet. The new version of the container image serves /metrics endpoint on a different port. ([#97621](https://github.com/kubernetes/kubernetes/pull/97621), [@vbannai](https://github.com/vbannai)) [SIG Cloud Provider] -- Use force unmount for NFS volumes if regular mount fails after 1 minute timeout ([#96844](https://github.com/kubernetes/kubernetes/pull/96844), [@gnufied](https://github.com/gnufied)) [SIG Storage] -- Users will see increase in time for deletion of pods and also guarantee that removal of pod from api server would mean deletion of all the resources from container runtime. ([#92817](https://github.com/kubernetes/kubernetes/pull/92817), [@kmala](https://github.com/kmala)) [SIG Node] -- Using exec auth plugins with kubectl no longer results in warnings about constructing many client instances from the same exec auth config. ([#97857](https://github.com/kubernetes/kubernetes/pull/97857), [@liggitt](https://github.com/liggitt)) [SIG API Machinery and Auth] -- Warning about using a deprecated volume plugin is logged only once. ([#96751](https://github.com/kubernetes/kubernetes/pull/96751), [@jsafrane](https://github.com/jsafrane)) [SIG Storage] - -### Other (Cleanup or Flake) - -- Bump github.com/Azure/go-autorest/autorest to v0.11.12 ([#97033](https://github.com/kubernetes/kubernetes/pull/97033), [@patrickshan](https://github.com/patrickshan)) [SIG API Machinery, CLI, Cloud Provider and Cluster Lifecycle] -- Delete deprecated mixed protocol annotation ([#97096](https://github.com/kubernetes/kubernetes/pull/97096), [@nilo19](https://github.com/nilo19)) [SIG Cloud Provider] -- Kube-proxy: Traffic from the cluster directed to ExternalIPs is always sent directly to the Service. ([#96296](https://github.com/kubernetes/kubernetes/pull/96296), [@aojea](https://github.com/aojea)) [SIG Network and Testing] -- Kubeadm: fix a whitespace issue in the output of the "kubeadm join" command shown as the output of "kubeadm init" and "kubeadm token create --print-join-command" ([#97413](https://github.com/kubernetes/kubernetes/pull/97413), [@SataQiu](https://github.com/SataQiu)) [SIG Cluster Lifecycle] -- Kubeadm: improve the error messaging when the user provides an invalid discovery token CA certificate hash. ([#97290](https://github.com/kubernetes/kubernetes/pull/97290), [@neolit123](https://github.com/neolit123)) [SIG Cluster Lifecycle] -- Migrate log messages in pkg/scheduler/{scheduler.go,factory.go} to structured logging ([#97509](https://github.com/kubernetes/kubernetes/pull/97509), [@aldudko](https://github.com/aldudko)) [SIG Scheduling] -- Migrate proxy/iptables/proxier.go logs to structured logging ([#97678](https://github.com/kubernetes/kubernetes/pull/97678), [@JornShen](https://github.com/JornShen)) [SIG Network] -- Migrate some scheduler log messages to structured logging ([#97349](https://github.com/kubernetes/kubernetes/pull/97349), [@aldudko](https://github.com/aldudko)) [SIG Scheduling] -- NONE ([#97167](https://github.com/kubernetes/kubernetes/pull/97167), [@geegeea](https://github.com/geegeea)) [SIG Node] -- NetworkPolicy validation framework optimizations for rapidly verifying CNI's work correctly across several pods and namespaces ([#91592](https://github.com/kubernetes/kubernetes/pull/91592), [@jayunit100](https://github.com/jayunit100)) [SIG Network, Storage and Testing] -- Official support to build kubernetes with docker-machine / remote docker is removed. This change does not affect building kubernetes with docker locally. ([#97618](https://github.com/kubernetes/kubernetes/pull/97618), [@jherrera123](https://github.com/jherrera123)) [SIG Release and Testing] -- Scheduler plugin validation now provides all errors detected instead of the first one. ([#96745](https://github.com/kubernetes/kubernetes/pull/96745), [@lingsamuel](https://github.com/lingsamuel)) [SIG Node, Scheduling and Testing] -- Storage related e2e testsuite redesign & cleanup ([#96573](https://github.com/kubernetes/kubernetes/pull/96573), [@Jiawei0227](https://github.com/Jiawei0227)) [SIG Storage and Testing] -- The OIDC authenticator no longer waits 10 seconds before attempting to fetch the metadata required to verify tokens. ([#97693](https://github.com/kubernetes/kubernetes/pull/97693), [@enj](https://github.com/enj)) [SIG API Machinery and Auth] -- The `AttachVolumeLimit` feature gate that is GA since v1.17 is now removed. ([#96539](https://github.com/kubernetes/kubernetes/pull/96539), [@ialidzhikov](https://github.com/ialidzhikov)) [SIG Storage] -- The `CSINodeInfo` feature gate that is GA since v1.17 is unconditionally enabled, and can no longer be specified via the `--feature-gates` argument. ([#96561](https://github.com/kubernetes/kubernetes/pull/96561), [@ialidzhikov](https://github.com/ialidzhikov)) [SIG Apps, Auth, Scheduling, Storage and Testing] -- The deprecated feature gates `RotateKubeletClientCertificate`, `AttachVolumeLimit`, `VolumePVCDataSource` and `EvenPodsSpread` are now unconditionally enabled and can no longer be specified in component invocations. ([#97306](https://github.com/kubernetes/kubernetes/pull/97306), [@gavinfish](https://github.com/gavinfish)) [SIG Node, Scheduling and Storage] -- `ServiceNodeExclusion`, `NodeDisruptionExclusion` and `LegacyNodeRoleBehavior`(locked to false) features have been promoted to GA. - To prevent control plane nodes being added to load balancers automatically, upgrade users need to add "node.kubernetes.io/exclude-from-external-load-balancers" label to control plane nodes. ([#97543](https://github.com/kubernetes/kubernetes/pull/97543), [@pacoxu](https://github.com/pacoxu)) [SIG API Machinery, Apps, Cloud Provider and Network] - -### Uncategorized - -- Adding Brazilian Portuguese translation for kubectl ([#61595](https://github.com/kubernetes/kubernetes/pull/61595), [@cpanato](https://github.com/cpanato)) [SIG CLI] - -## Dependencies - -### Added -_Nothing has changed._ - -### Changed -- github.com/Azure/go-autorest/autorest: [v0.11.1 → v0.11.12](https://github.com/Azure/go-autorest/autorest/compare/v0.11.1...v0.11.12) -- github.com/coredns/corefile-migration: [v1.0.10 → v1.0.11](https://github.com/coredns/corefile-migration/compare/v1.0.10...v1.0.11) -- github.com/golang/mock: [v1.4.1 → v1.4.4](https://github.com/golang/mock/compare/v1.4.1...v1.4.4) -- github.com/google/cadvisor: [v0.38.5 → v0.38.6](https://github.com/google/cadvisor/compare/v0.38.5...v0.38.6) -- github.com/heketi/heketi: [c2e2a4a → v10.2.0+incompatible](https://github.com/heketi/heketi/compare/c2e2a4a...v10.2.0) -- github.com/miekg/dns: [v1.1.4 → v1.1.35](https://github.com/miekg/dns/compare/v1.1.4...v1.1.35) -- k8s.io/system-validators: v1.2.0 → v1.3.0 - -### Removed -- rsc.io/quote/v3: v3.1.0 -- rsc.io/sampler: v1.3.0 diff --git a/content/ko/releases/_index.md b/content/ko/releases/_index.md new file mode 100644 index 0000000000..aa6a306f8a --- /dev/null +++ b/content/ko/releases/_index.md @@ -0,0 +1,27 @@ +--- +linktitle: 릴리스 히스토리 +title: 릴리스 +type: docs +--- + + + + +쿠버네티스 프로젝트는 가장 최신의 3개 마이너(minor) 릴리스({{< skew latestVersion >}}, {{< skew prevMinorVersion >}}, {{< skew oldestMinorVersion >}})에 대해서 릴리스 브랜치를 관리한다. 쿠버네티스 1.19 및 이후 신규 버전은 약 1년간 패치 지원을 받을 수 있다. 쿠버네티스 1.18 및 이전 버전은 약 9개월간의 패치 지원을 받을 수 있다. + +쿠버네티스 버전은 **x.y.z** 의 형태로 표현되는데, +**x** 는 메이저(major) 버전, **y** 는 마이너(minor), **z** 는 패치(patch) 버전을 의미하며, 이는 [시맨틱 버전](https://semver.org/)의 용어를 따른 것이다. + +저 자세한 정보는 [버전 차이(skew) 정책](/releases/version-skew-policy/) 문서에서 확인하길 바란다. + + + +## 릴리스 히스토리 + +{{< release-data >}} + +## 차기 릴리스 + +차기 쿠버네티스 릴리스 **{{< skew nextMinorVersion >}}** 일정은 [스케줄](https://github.com/kubernetes/sig-release/tree/master/releases/release-{{< skew nextMinorVersion >}})에서 확인할 수 있다. + +## 유용한 자원 diff --git a/content/ko/releases/notes.md b/content/ko/releases/notes.md new file mode 100644 index 0000000000..b509ae0d65 --- /dev/null +++ b/content/ko/releases/notes.md @@ -0,0 +1,13 @@ +--- +linktitle: 릴리스 노트 +title: 노트 +type: docs +description: > + 쿠버네티스 릴리스 노트. +sitemap: + priority: 0.5 +--- + +릴리스 노트는 사용자의 쿠버네티스 버전에 해당하는 [변경로그(Changelog)](https://github.com/kubernetes/kubernetes/tree/master/CHANGELOG)를 통해서 확인할 수 있다. {{< skew latestVersion >}} 의 변경로그는 [깃허브](https://github.com/kubernetes/kubernetes/blob/master/CHANGELOG/CHANGELOG-{{< skew latestVersion >}}.md)에 있다. + +대안으로, 릴리스 노트는 [relnotes.k8s.io](https://relnotes.k8s.io)에서 온라인으로 검색 및 필터링이 가능하다. {{< skew latestVersion >}}로 필터링된 릴리스 노트는 [relnotes.k8s.io](https://relnotes.k8s.io/?releaseVersions={{< skew latestVersion >}}.0)에서 확인한다. diff --git a/content/ko/docs/setup/release/version-skew-policy.md b/content/ko/releases/version-skew-policy.md similarity index 97% rename from content/ko/docs/setup/release/version-skew-policy.md rename to content/ko/releases/version-skew-policy.md index 76ff7504fd..38052aa18d 100644 --- a/content/ko/docs/setup/release/version-skew-policy.md +++ b/content/ko/releases/version-skew-policy.md @@ -20,8 +20,8 @@ weight: 30 ## 지원되는 버전 -쿠버네티스 버전은 **x.y.z**로 표현되는데, -여기서 **x**는 메이저 버전, **y**는 마이너 버전, **z**는 [시맨틱 버전](https://semver.org/) 용어에 따른 패치 버전이다. +쿠버네티스 버전은 **x.y.z** 로 표현되는데, +여기서 **x** 는 메이저 버전, **y** 는 마이너 버전, **z** 는 패치 버전을 의미하며, 이는 [시맨틱 버전](https://semver.org/) 용어에 따른 것이다. 자세한 내용은 [쿠버네티스 릴리스 버전](https://github.com/kubernetes/community/blob/master/contributors/design-proposals/release/versioning.md#kubernetes-release-versioning)을 참조한다. 쿠버네티스 프로젝트는 최근 세 개의 마이너 릴리스 ({{< skew latestVersion >}}, {{< skew prevMinorVersion >}}, {{< skew oldestMinorVersion >}}) 에 대한 릴리스 분기를 유지한다. 쿠버네티스 1.19 이상은 약 1년간의 패치 지원을 받는다. 쿠버네티스 1.18 이상은 약 9개월의 패치 지원을 받는다. From 7ebb2271223df37a414b117921b3da78428ee645 Mon Sep 17 00:00:00 2001 From: Jihoon Seo Date: Mon, 31 May 2021 18:23:51 +0900 Subject: [PATCH 05/10] [ko] Remove untranslated files in ko/case-studies/ --- .../adform/adform_featured_logo.png | Bin 7830 -> 0 bytes content/ko/case-studies/adform/index.html | 86 ---------------- .../case-studies/amadeus/amadeus_featured.png | Bin 6970 -> 0 bytes .../ko/case-studies/amadeus/amadeus_logo.png | Bin 19562 -> 0 bytes content/ko/case-studies/amadeus/index.html | 84 ---------------- .../ancestry/ancestry_featured.png | Bin 21016 -> 0 bytes .../case-studies/ancestry/ancestry_logo.png | Bin 21141 -> 0 bytes content/ko/case-studies/ancestry/index.html | 92 ----------------- .../blablacar/blablacar_featured.png | Bin 7867 -> 0 bytes .../case-studies/blablacar/blablacar_logo.png | Bin 7841 -> 0 bytes content/ko/case-studies/blablacar/index.html | 85 ---------------- .../blackrock/blackrock_featured.png | Bin 6391 -> 0 bytes .../case-studies/blackrock/blackrock_logo.png | Bin 6430 -> 0 bytes content/ko/case-studies/blackrock/index.html | 83 ---------------- .../case-studies/buffer/buffer_featured.png | Bin 5263 -> 0 bytes .../ko/case-studies/buffer/buffer_logo.png | Bin 4791 -> 0 bytes content/ko/case-studies/buffer/index.html | 83 ---------------- .../capital-one/capitalone_featured_logo.png | Bin 11306 -> 0 bytes .../ko/case-studies/capital-one/index.html | 61 ------------ .../crowdfire/crowdfire_featured_logo.png | Bin 7827 -> 0 bytes content/ko/case-studies/crowdfire/index.html | 85 ---------------- .../case-studies/golfnow/golfnow_featured.png | Bin 22915 -> 0 bytes .../ko/case-studies/golfnow/golfnow_logo.png | Bin 8858 -> 0 bytes content/ko/case-studies/golfnow/index.html | 89 ----------------- .../haufegroup/haufegroup_featured.png | Bin 5841 -> 0 bytes .../haufegroup/haufegroup_logo.png | Bin 5794 -> 0 bytes content/ko/case-studies/haufegroup/index.html | 85 ---------------- .../case-studies/huawei/huawei_featured.png | Bin 14310 -> 0 bytes .../ko/case-studies/huawei/huawei_logo.png | Bin 14274 -> 0 bytes content/ko/case-studies/huawei/index.html | 73 -------------- .../ko/case-studies/ibm/ibm_featured_logo.png | Bin 11298 -> 0 bytes .../ko/case-studies/ibm/ibm_featured_logo.svg | 1 - content/ko/case-studies/ibm/index.html | 80 --------------- content/ko/case-studies/ing/index.html | 78 --------------- .../ko/case-studies/ing/ing_featured_logo.png | Bin 10305 -> 0 bytes content/ko/case-studies/naic/index.html | 87 ---------------- .../case-studies/naic/naic_featured_logo.png | Bin 9433 -> 0 bytes .../ko/case-studies/newyorktimes/index.html | 74 -------------- .../newyorktimes/newyorktimes_featured.png | Bin 20189 -> 0 bytes .../newyorktimes/newyorktimes_logo.png | Bin 10251 -> 0 bytes content/ko/case-studies/nordstrom/index.html | 75 -------------- .../nordstrom/nordstrom_featured_logo.png | Bin 7352 -> 0 bytes .../northwestern-mutual/index.html | 69 ------------- .../northwestern_featured_logo.png | Bin 10083 -> 0 bytes content/ko/case-studies/ocado/index.html | 83 ---------------- .../ocado/ocado_featured_logo.png | Bin 8089 -> 0 bytes content/ko/case-studies/openAI/index.html | 69 ------------- .../case-studies/openAI/openai_featured.png | Bin 12132 -> 0 bytes .../ko/case-studies/openAI/openai_logo.png | Bin 19818 -> 0 bytes content/ko/case-studies/peardeck/index.html | 87 ---------------- .../peardeck/peardeck_featured.png | Bin 11786 -> 0 bytes .../case-studies/peardeck/peardeck_logo.png | Bin 9260 -> 0 bytes content/ko/case-studies/pearson/index.html | 83 ---------------- .../case-studies/pearson/pearson_featured.png | Bin 7784 -> 0 bytes .../ko/case-studies/pearson/pearson_logo.png | Bin 5123 -> 0 bytes content/ko/case-studies/pinterest/index.html | 84 ---------------- .../pinterest/pinterest_feature.png | Bin 9118 -> 0 bytes .../case-studies/pinterest/pinterest_logo.png | Bin 9003 -> 0 bytes content/ko/case-studies/slingtv/index.html | 79 --------------- .../slingtv/slingtv_featured_logo.png | Bin 8513 -> 0 bytes .../ko/case-studies/squarespace/index.html | 71 ------------- .../squarespace/squarespace_featured_logo.png | Bin 4539 -> 0 bytes content/ko/case-studies/wikimedia/index.html | 66 ------------- .../wikimedia/wikimedia_featured.png | Bin 21322 -> 0 bytes .../case-studies/wikimedia/wikimedia_logo.png | Bin 7875 -> 0 bytes content/ko/case-studies/wink/index.html | 87 ---------------- .../ko/case-studies/wink/wink_featured.png | Bin 21292 -> 0 bytes content/ko/case-studies/wink/wink_logo.png | Bin 5623 -> 0 bytes content/ko/case-studies/workiva/index.html | 93 ------------------ .../workiva/workiva_featured_logo.png | Bin 5980 -> 0 bytes .../ko/case-studies/yahoo-japan/index.html | 4 - .../yahoo-japan/yahooJapan_logo.png | Bin 7266 -> 0 bytes content/ko/case-studies/ygrene/index.html | 82 --------------- .../ygrene/ygrene_featured_logo.png | Bin 11569 -> 0 bytes content/ko/case-studies/zalando/index.html | 83 ---------------- .../zalando/zalando_feature_logo.png | Bin 7643 -> 0 bytes 76 files changed, 2341 deletions(-) delete mode 100644 content/ko/case-studies/adform/adform_featured_logo.png delete mode 100644 content/ko/case-studies/adform/index.html delete mode 100644 content/ko/case-studies/amadeus/amadeus_featured.png delete mode 100644 content/ko/case-studies/amadeus/amadeus_logo.png delete mode 100644 content/ko/case-studies/amadeus/index.html delete mode 100644 content/ko/case-studies/ancestry/ancestry_featured.png delete mode 100644 content/ko/case-studies/ancestry/ancestry_logo.png delete mode 100644 content/ko/case-studies/ancestry/index.html delete mode 100644 content/ko/case-studies/blablacar/blablacar_featured.png delete mode 100644 content/ko/case-studies/blablacar/blablacar_logo.png delete mode 100644 content/ko/case-studies/blablacar/index.html delete mode 100644 content/ko/case-studies/blackrock/blackrock_featured.png delete mode 100644 content/ko/case-studies/blackrock/blackrock_logo.png delete mode 100644 content/ko/case-studies/blackrock/index.html delete mode 100644 content/ko/case-studies/buffer/buffer_featured.png delete mode 100644 content/ko/case-studies/buffer/buffer_logo.png delete mode 100644 content/ko/case-studies/buffer/index.html delete mode 100644 content/ko/case-studies/capital-one/capitalone_featured_logo.png delete mode 100644 content/ko/case-studies/capital-one/index.html delete mode 100644 content/ko/case-studies/crowdfire/crowdfire_featured_logo.png delete mode 100644 content/ko/case-studies/crowdfire/index.html delete mode 100644 content/ko/case-studies/golfnow/golfnow_featured.png delete mode 100644 content/ko/case-studies/golfnow/golfnow_logo.png delete mode 100644 content/ko/case-studies/golfnow/index.html delete mode 100644 content/ko/case-studies/haufegroup/haufegroup_featured.png delete mode 100644 content/ko/case-studies/haufegroup/haufegroup_logo.png delete mode 100644 content/ko/case-studies/haufegroup/index.html delete mode 100644 content/ko/case-studies/huawei/huawei_featured.png delete mode 100644 content/ko/case-studies/huawei/huawei_logo.png delete mode 100644 content/ko/case-studies/huawei/index.html delete mode 100644 content/ko/case-studies/ibm/ibm_featured_logo.png delete mode 100644 content/ko/case-studies/ibm/ibm_featured_logo.svg delete mode 100644 content/ko/case-studies/ibm/index.html delete mode 100644 content/ko/case-studies/ing/index.html delete mode 100644 content/ko/case-studies/ing/ing_featured_logo.png delete mode 100644 content/ko/case-studies/naic/index.html delete mode 100644 content/ko/case-studies/naic/naic_featured_logo.png delete mode 100644 content/ko/case-studies/newyorktimes/index.html delete mode 100644 content/ko/case-studies/newyorktimes/newyorktimes_featured.png delete mode 100644 content/ko/case-studies/newyorktimes/newyorktimes_logo.png delete mode 100644 content/ko/case-studies/nordstrom/index.html delete mode 100644 content/ko/case-studies/nordstrom/nordstrom_featured_logo.png delete mode 100644 content/ko/case-studies/northwestern-mutual/index.html delete mode 100644 content/ko/case-studies/northwestern-mutual/northwestern_featured_logo.png delete mode 100644 content/ko/case-studies/ocado/index.html delete mode 100644 content/ko/case-studies/ocado/ocado_featured_logo.png delete mode 100644 content/ko/case-studies/openAI/index.html delete mode 100644 content/ko/case-studies/openAI/openai_featured.png delete mode 100644 content/ko/case-studies/openAI/openai_logo.png delete mode 100644 content/ko/case-studies/peardeck/index.html delete mode 100644 content/ko/case-studies/peardeck/peardeck_featured.png delete mode 100644 content/ko/case-studies/peardeck/peardeck_logo.png delete mode 100644 content/ko/case-studies/pearson/index.html delete mode 100644 content/ko/case-studies/pearson/pearson_featured.png delete mode 100644 content/ko/case-studies/pearson/pearson_logo.png delete mode 100644 content/ko/case-studies/pinterest/index.html delete mode 100644 content/ko/case-studies/pinterest/pinterest_feature.png delete mode 100644 content/ko/case-studies/pinterest/pinterest_logo.png delete mode 100644 content/ko/case-studies/slingtv/index.html delete mode 100644 content/ko/case-studies/slingtv/slingtv_featured_logo.png delete mode 100644 content/ko/case-studies/squarespace/index.html delete mode 100644 content/ko/case-studies/squarespace/squarespace_featured_logo.png delete mode 100644 content/ko/case-studies/wikimedia/index.html delete mode 100644 content/ko/case-studies/wikimedia/wikimedia_featured.png delete mode 100644 content/ko/case-studies/wikimedia/wikimedia_logo.png delete mode 100644 content/ko/case-studies/wink/index.html delete mode 100644 content/ko/case-studies/wink/wink_featured.png delete mode 100644 content/ko/case-studies/wink/wink_logo.png delete mode 100644 content/ko/case-studies/workiva/index.html delete mode 100644 content/ko/case-studies/workiva/workiva_featured_logo.png delete mode 100644 content/ko/case-studies/yahoo-japan/index.html delete mode 100644 content/ko/case-studies/yahoo-japan/yahooJapan_logo.png delete mode 100644 content/ko/case-studies/ygrene/index.html delete mode 100644 content/ko/case-studies/ygrene/ygrene_featured_logo.png delete mode 100644 content/ko/case-studies/zalando/index.html delete mode 100644 content/ko/case-studies/zalando/zalando_feature_logo.png diff --git a/content/ko/case-studies/adform/adform_featured_logo.png b/content/ko/case-studies/adform/adform_featured_logo.png deleted file mode 100644 index 7e3be727e39e676c4039d3d7e4fa5fff3de0ebe9..0000000000000000000000000000000000000000 GIT binary patch literal 0 HcmV?d00001 literal 7830 zcmc(EcUY6n(l=E)(t8bEfzV5&69Nc9dhwAGAOa#u2mt~jy-E`hDJoUzRiroRO*)8D zl-{dUsb74~dCvR&ajy5@cdu)+_r2H5?Ci|kv%lFy85(F)k+YHG;o(u~>VS=L;}ot3 zkP_g24-G%Va04qw%MxRPe2npiq2YMyj!1hrkFFcc32qFBIr_Nw!Ikjv2!<Eisl* zeUJmvO&s=zMjY#g!g1r_DXC&nFb7vShQ}W6@~Imcw8Kn0T!}Q2`EYv?(CxD zi-w!{8kjoxx;iL20#sFal&~P205>=W#)EZxitqqol>vXt1>v6mOalQtf0JNbl>z@! z$`WeGqlrYrd1S@q#2h4LBzWW%#U$3l#l{SNr~Yo#5{Zu7#LOz z;lcNh1Tfsg0qugqxF8Wce~>Ed@m>1!*m5SxpE;R$4|wTH_yHU4#b)hH!xY z!|U=tyc+*2FGv#&hhdOtQzY`~Ki)HRMq-d2&PWuGrsm(?;t__z99$58CPn@j_}}`% z(Jo$aM+h3}#`Cv@Kra793n>XLMFp^&l%l$<%zxj(zjz)0%kCw#B($XEArgw}^3wm6 z7l=~^_{Y8fpS%B=!o|>^$NvN`?&Uw>4@cl)9*qm`%NG4ZJUn(HU9h?-c6!^I3}b4R z-g4S(p_kZCpc@&bM%T_nlMQEl6lhMAo|o@w`H<5^O<@G&_Ta&|_66p_vn(eCr@R>t zU(92$QDbsmh_;$KE$w#>Dgin5&CShMF(3~#`uuo{`69Zg zNKRJvooYbKr%#`9a@rSA?>~HiKYmQjt>V}Fl6PZcBa#1nS)S0>_3B4d*%mP|@m}kV z3QJOQa`oduwhE!w%x%+G7pKUdEpvOP10O}hvLf)jy1Tl>Y+EjO8~2#N3{3SN42+CB zv-RVSQfY&$zkdA^`>RGx*6Z9H4qutBc9lo|h;gN%E)R)(WqjG?>F1{ud3N#h=DIh5 zFGjMBtgYPuPmlZr6-x||lacw{3km2#)T-WG9!^e9%Dtfih96SuE&z8?;u-^Ta&qJb zq)c7}&gJFFEz*Q+q(pL0zs-9!mYNs0sRodq^)DT$QLuqa%FB^1F0^lHMs=MKPj#M! z1;o_R2er4iFD}|gceSeqyX}WmRaHe$vF8cDnA++h90|zcNlnkp{P_WuPgnsdnxfgc zGrASE3-}TXpv$8x^a^y;j^~m!6(pdoK=maGnB?i>6K9)e zlQN4x!d{*oG(S5P6cQ@8tRygR-1<#a;2U)2HLfN(-XHlxO~$Yy$`InrkzibqsJT2t zO32x4df)NBxTI?DtIE)a#5XNhqD{WXo5^=A)D+5yVZm#IZ{%ISmXIDRI&@R+)7j)@ zOF5W*@#Q>d(Nz8`?cx0+P#y^>U7ov)KYrDaK);U0g=CIIPM8um)%}Fv^b%H%WWRf< z6PamwG*9S7FXgIe47ik3`H0jpN-ksXPnbAlp!)jT_uJXvRGn8+F5h)NxyG#XGRvlQ zL*X^~Q7$j28Cl+2lay%`JKfpRXMJF$Jb9lKYLAvfkTLnY(sI#8kOsv`5BY20Z z^^2*_{pJ~=*<=C3bQxnS3z7pSfQc{8f(cS0t_1n4TJja-O#b&ozJ)9W#1y38q<;84 z)o_Z64A?(psUC8yal#yi%$Yxq5$u3}_%cv%UTQ)9JRvO7kI&4|J0BvmdhA=nFH=Sh z@=1)6s0(Cza-seBLNT{c-r6{76ss}uvMVn*%sPBvrKr5z8}zQiTZDJtdH(VI;cqrv zV`rz8Jlgx&I+H6zE0{W*g5l$AwHbc+RZzB-GJJhPKQ&WF_tQFm6!;ANBok9efAm>AXK=99@; zggjE^&0shK9UWbcYI)Hesp^@X_2I(8!d`iMqElqQ2E&K<@835yH8nIeR%)~JID~%v6aLsIe3@ZL zgi!U=GfB_H>FH@{r=drUUZpuX1apj)0mjB-pMwacrKNG-5^sTlKz!EDVplYlb5Y}o zvq~tUHY~7)T^_;lp82337tHO?{|K* zf%|G+zkdBB{T6MmeAYA%{!QdB^`KZb*(cYRzlGs3eZc|GPPT(k2`cjMbm$9%AWG4d z4m~R8>$yS?GDL-iKU?$?k8vxamy67bd8nJvXZs6NwgH^-K6}e6Q4;-cU(p_7{T$s- zw!cR+1LIgq4xB9}&m%4hg*=NYviJ&&MbaqG^i#rXZu@7G9vc?DVCWNkUhR>2d*Js z^@dR}|9H07td}CLPK>4x0%OT_$T6X&+&;^+BMGOItE;OIRIiS3Hk{{9VNbaOl17Rg zEr^)peMy(StDtLEvKRLOVunJ#DQJ)HFP zDDrilHAWE~u;@!mg&Ed~Do|LI+fa=yt>O-VQ7iRaM5;A^+(gE+-THfbO^4^uRYb{~ zq(ZJq1>z`!R0s*jWbz)Saa_K58O3Skq0tql3rKT6+DI6?44R+|X?;O=y#fsQ-Ws6l zIFLf@DeQyy^lUWVU&f(3hSX?YYky;>_UEmeoM^U&{mLG$H;NvD8GZ*{%&q5}ca>(d z^*W{m7Q`ZVdsMI8ArOdsC*WpPzBS64qRhIH={|+nV6{=G)w1{56GFT%3g~50j3q-) zT=~RwgXdjwg;Ku674G^cq1}sD!&v53&$0Kq>n##3xX{sv-^_T^F1ytkML$|PgLz|P zcC-BXvvy26i-lYIeG~bORaSX=wLmw`r;f%An4%!r-ytnayLWCCE2|_wTrpU@s{P?h zF*j*RgW>D*=<>QebH|M06W$6HBA|8qR)PN$kbp8`N6Ud@onE|nF%sN+fb%JBT6n9y z<3lsE<1S{~S&w05YtBod%JRo}-))~Cn_=mk(YZ2pQQgu^MEkCq;=)v1XG`2Sy|GUt zzE!(2f8xoI`PRXFd4NH2W;~8~o)ADhuIhW_jCxfJ`*H7}R(T_qgmRwxl3#}95yw&= zdhKR*bw;D0P?n4-D=Hc@F#E>#hSx|cTh+(?W1>CTn4&r_(t7MG^Hyw-Gm=vYOA|^u zb@a&P0ECWWJzlx{RnBd?^G!G**TmiOrkqeaoepOee>c&jB{qD;cG^%NBIc4E+EHGA zyW19Da=r)X)aY5 z$!$6_=V60a9-}#072Bn|t*7q`555ys<@}mWWqa$S_vnGM#)DXBjE*A^YRtrKyu>&* z4E`xPO2=@%?fm35D9&-zg6e{hSIw}1A~YTwT!qgcQcU8CC$b=^;^IF??~J$}&%6$O z(_M%EquqfNOk=3AszZ}=S6uuS^MT~-4aF3-(k@sc^hC=pG%tvJS#gmu)O&v@n;Mro zX77rEkFog+zuOdYWa3q?u{Jg~SSzC$dLe_)URfh=(}h_Jy!M(;Azu=+Ywme%*AcjR>2&Vl!lTPuXnt==C2~+J|MIWU-B`B_$30aNgYD#Q zbJ_8}qMyDEA`4p;X8?RXIZ#ZeH<0W<8q)(FAR^W{sYz_WsdQ!3>a;pkUnIbRM^hzM zgt}VqqwYnvz7+KHppopCAi~BYp7@1=u~~yFq~h(GZkhf>L$ZvsW!x5yA0|^b?ik&w zs3TTxVWR(Nsr=~1BkH~6k!9l3oEtv?YWP%NE=>zFRpm6y=K3w`6B>>7KOfJ`d;dOE zH9XRB`jd;ZAb127G?J9f6T?6=&cYTA%)7)TCbA%H%3o`1Ye4G;2DgYIj$IgESfNQi zU@?m2OE8yKPj@u8-#&hAQe9bbafmqt-}eWQs5!%TQ-sCb#_*_0SQ9)iT+0xigwGUy zAxyKfF`n;J(o?N`Ki*lMTnh(qRZLlc-6|p^?`W?jYDM0G?b4gOf7^$bR8`P~PQO0w zB+@0(sZHFw{p6_;)UeK7%7wf_UXQQX!o-A-v5UCtIsB-Pl$1tn<6HNY&*npD8Ds)atedcs*r%QISPHMzD4~pXloDg#P^Kz343M~zftUVkxi6`i z?j7EG@H$nod8zMuc0|swlbRz+p>D0s%+$@zO@Yu*mxWYQeB?ypJH*d7SHi13l&rqK zzQAn{Dq#KOdnS#7Yx1j_EU0=1;pI@+<#E5iV88k$&B$2oElE4;P8dZ&vhZEt0b`8pmsduCIJp1A zp4Z|6S4g~z3uigNpSrJEAt1zi(?oCC;jX;{H{Z&ZE@Bd;eJoX zBS_`el>qhW?sc1e#E0@2J2Yg^%o|>33i?M}=-D7~Y9j zDoTlX$n~vQ;LzomEbsAZeHr8C)t6#oi=Qu5LJ%Km=7{l0Y^_29NqH+IYX{|-=Fq{0 zo)%$S^9sZ*59?$rrP3#gf1hI_^k+W}l6|%{qbQ45H(95qfVs6bJ-;~NWmwD=+gbYf z7q!aVh7spPiJx0~>fX|}tVGVe%N22`hRQA4km80kJDBOkv;Fxt>qaj#GqWI{g#>1o zH|7gV+jBO?=g@2cLrzR_niXHeY(MGCPCWtFJtGaLpvCUne$(hE8}7!+-sgyBPi9Nf zweqYiyVKn{KO1o+^Of8}zYKQB_s3O1VTjrbCIH}mk>ui7b)dlvWe&rmA0KSio98>r zrd%-okSG(s0HPaKbjnRsP~6iNS7yMzFMrVto#9wosqQIRUAf9 zabN5d>(r=v8~IUA;(Ny{?4pZ_erOS6%+Xj2zxl$CoqekPjWlD_mWiQdMHl@ze|@}& zrT~N-@ay^MoHs7P#)kF#={w_4<#_&s;lV+uuI|Es&6yG~v$`paXn}m9d!V!IynOwB z5hcZUIc!d{adX`9k#Ts|vySvr+vi_P%e;z>-)^sDlU+lH*FD&??CrCM*JkS69qJe% zU?t~~ECIdOYByRXSOK@ALGNmRQ|x@^J*^ zEYi5SNT{^cytxW3!Z1&~Ce_;gKIG$?1wIE&Pj=nPpZ?o$cP(;|1Y; z=abTv=!XO;u1y|#4_7_3BHwcOpPbPs-Yo-i39_5egE*)iN)>H&*{nT-gzb23QQaw0 z`;~9DUT7P&%imUOZD03RDO)(5nwSVhPHs$<+tzw>*Nlg`E7q`P7qEO7J84j2OrK;p z*XK`@DNP~B{NX1K4I0>crRq0%r6{mrJ8|>1T;!FrGhzJDxmE9&QXw$o3k@H|`1rWu zj>3m_d5>Arwc*UQt!>4=^V2=y zF-@XtTzxjhdUnSyxv-DOvtTMIFCg)9FM|5j2HV@9Uz*jeKYZXmxqJ3SVUL`Baw=fS zfNJq6xBW^1pI@gWCq`N+c_0z ze2>;g4J*5^?UinZ?(K0Xpm_Xx4m>r#PK3%lcn5-iv*o{sSsz&+sA>#q>fEE!uhXL{ zEvz(L9P!&zZ02qpG?=qb6#Zz^vB{mxW?{5r|I|qO(A7+6^Cmaz>B(q=s>CW(D&hX{ z_D%Cixr&$#K%J?kVEG5-26HaFYQj6fhDXpt&awC9r%y8e*V485PD5{fa25>M58N>hWOu@l@+|az0sHuqCqTpnIVs z;@}eJYafo-DV&KLpI>;@;V<-<g7{z!S;aZ^Wm3X&$^aCuCEK8=9%(M)7Eai z9mZI^mHp|cn}_&*$aMI;SM+r2Krc<|wXR5ZZp~R`nlEY|h<)lBGVXn@zw{cgrjjy^ zckJ>yn7&i?2kS+7%eVDmDZO30_c@89eV91~fw{5#fCYdcQ_k-cplu6c4&>lXR}k`Y zrFzOHBxA|Lw<1+Mg+ia)_IubHiI&^SM~2TD0%5n6_SImpg|293pU%O|delNYuB1F* zY9-t3DDAZ9b{6<44RRXdaEP}WwvTi>>HPGQv-u>3=vJa#nq&Xl^t~je8_7PIiS$&l z!mAagHw|;!KjI8y)5B?9Buv6xjf*TwF}LpAQA!#1xxek7VvrUroMGI_^O4`8VU4ro z>Omv2LJPeWgY}eZ$UH*+Ob8b^Ip(&CiVCx%oUPmQKW!ZGh>D8xqZ>Z^&7VKsXASZ1 z!G_(WcSsK_5h_o^K4wkhE*_@JtSOMgEawTD`{AEEU>(_=vX<23ak!faTrDzJ;3k@z zXy`mXX7SZ$K<_Qkl0P;quIH6VSAfkb5JN&vCTr3sXjMWd?k9EEw2#@}`c!K1aVIyZ zo9!UBeVCjtUDqC@_q)ht!^v}N_DPc&`T|2Y(if6;R4-|4FcEvQlOpNu0Li*Kzc83G zh@f~8-Aq5cv_>uDod}E#qh3a`nzgz;DL9> ze7~v~mfyFyq9D25Hng`IGHAiMZ*y{SpTO5DaKag#B)$cfGg6GZlP#h)C+Z@Y1|nFx zZ9P-5w@G0mHUzdFQbU!`c{nVq>c8}$1MS`Ty=<%79YMxXAEEp$Gk7D`wfpuxf4$@i zARgA{=ITDw8BvIj2kvY!GX3gTx;h#!wkQwhBVD4fp7akP*bRJec1Kl}`zT$~0g_+F z9!yz5{^IL5r4beAJ{L9MSJ*C5m4gx`ry%>az}WH;6bc>e?|)JGQBEsOexKRZR36}PuAG4R-OiLg^&%Qtj)O7-W?o_)NklQJMpefX=Q{Zj-Y`6SCzvS4jhn8iN zoVx35(AW-=cVtfM6k0P@ z&86RNeipR#iU#pykqp0)9%$)vIWG})vA+YYeJMjhyROA_!dWpwnmMNpl+pVz=&LJT z=#>OfbtB^mr{>DecmiWJ7ZXw+evn-+H}+GXZds)^+n~a5{&nbjK0$C8!JqzrG}PX@ deh`R{_fV%)x((o6_vfEtT`dD}nTB2P{{xeilN$g4 diff --git a/content/ko/case-studies/adform/index.html b/content/ko/case-studies/adform/index.html deleted file mode 100644 index 1de43d0637..0000000000 --- a/content/ko/case-studies/adform/index.html +++ /dev/null @@ -1,86 +0,0 @@ ---- -title: Adform Case Study -linkTitle: Adform -case_study_styles: true -cid: caseStudies -logo: adform_featured_logo.png -draft: false -featured: true -weight: 47 -quote: > - Kubernetes enabled the self-healing and immutable infrastructure. We can do faster releases, so our developers are really happy. They can ship our features faster than before, and that makes our clients happier. - -new_case_study_styles: true -heading_background: /images/case-studies/adform/banner1.jpg -heading_title_logo: /images/adform_logo.png -subheading: > - Improving Performance and Morale with Cloud Native -case_study_details: - - Company: AdForm - - Location: Copenhagen, Denmark - - Industry: Adtech ---- - -

Challenge

- -

Adform's mission is to provide a secure and transparent full stack of advertising technology to enable digital ads across devices. The company has a large infrastructure: OpenStack-based private clouds running on 1,100 physical servers in 7 data centers around the world, 3 of which were opened in the past year. With the company's growth, the infrastructure team felt that "our private cloud was not really flexible enough," says IT System Engineer Edgaras Apšega. "The biggest pain point is that our developers need to maintain their virtual machines, so rolling out technology and new software takes time. We were really struggling with our releases, and we didn't have self-healing infrastructure."

- -

Solution

- -

The team, which had already been using Prometheus for monitoring, embraced Kubernetes and cloud native practices in 2017. "To start our Kubernetes journey, we had to adapt all our software, so we had to choose newer frameworks," says Apšega. "We also adopted the microservices way, so observability is much better because you can inspect the bug or the services separately."

- -

Impact

- -

"Kubernetes helps our business a lot because our features are coming to market faster," says Apšega. The release process went from several hours to several minutes. Autoscaling has been at least 6 times faster than the semi-manual VM bootstrapping and application deployment required before. The team estimates that the company has experienced cost savings of 4-5x due to less hardware and fewer man hours needed to set up the hardware and virtual machines, metrics, and logging. Utilization of the hardware resources has been reduced as well, with containers notching 2-3 times more efficiency over virtual machines. "The deployments are very easy because developers just push the code and it automatically appears on Kubernetes," says Apšega. Prometheus has also had a positive impact: "It provides high availability for metrics and alerting. We monitor everything starting from hardware to applications. Having all the metrics in Grafana dashboards provides great insight on your systems."

- -{{< case-studies/quote author="Edgaras Apšega, IT Systems Engineer, Adform" >}} -"Kubernetes enabled the self-healing and immutable infrastructure. We can do faster releases, so our developers are really happy. They can ship our features faster than before, and that makes our clients happier." -{{< /case-studies/quote >}} - -{{< case-studies/lead >}} -Adform made headlines last year when it detected the HyphBot ad fraud network that was costing some businesses hundreds of thousands of dollars a day. -{{< /case-studies/lead >}} - -

With its mission to provide a secure and transparent full stack of advertising technology to enable an open internet, Adform published a white paper revealing what it did—and others could too—to limit customers' exposure to the scam.

- -

In that same spirit, Adform is sharing its cloud native journey. "When you see that everyone shares their best practices, it inspires you to contribute back to the project," says IT Systems Engineer Edgaras Apšega.

- -

The company has a large infrastructure: OpenStack-based private clouds running on 1,100 physical servers in their own seven data centers around the world, three of which were opened in the past year. With the company's growth, the infrastructure team felt that "our private cloud was not really flexible enough," says Apšega. "The biggest pain point is that our developers need to maintain their virtual machines, so rolling out technology and new software really takes time. We were really struggling with our releases, and we didn't have self-healing infrastructure."

- -{{< case-studies/quote - image="/images/case-studies/adform/banner3.jpg" - author="Edgaras Apšega, IT Systems Engineer, Adform" ->}} -"The fact that Cloud Native Computing Foundation incubated Kubernetes was a really big point for us because it was vendor neutral. And we can see that a community really gathers around it. Everyone shares their experiences, their knowledge, and the fact that it's open source, you can contribute." -{{< /case-studies/quote >}} - -

The team, which had already been using Prometheus for monitoring, embraced Kubernetes, microservices, and cloud native practices. "The fact that Cloud Native Computing Foundation incubated Kubernetes was a really big point for us because it was vendor neutral," says Apšega. "And we can see that a community really gathers around it."

- -

A proof of concept project was started, with a Kubernetes cluster running on bare metal in the data center. When developers saw how quickly containers could be spun up compared to the virtual machine process, "they wanted to ship their containers in production right away, and we were still doing proof of concept," says IT Systems Engineer Andrius Cibulskis.

- -

Of course, a lot of work still had to be done. "First of all, we had to learn Kubernetes, see all of the moving parts, how they glue together," says Apšega. "Second of all, the whole CI/CD part had to be redone, and our DevOps team had to invest more man hours to implement it. And third is that developers had to rewrite the code, and they're still doing it."

- -

The first production cluster was launched in the spring of 2018, and is now up to 20 physical machines dedicated for pods throughout three data centers, with plans for separate clusters in the other four data centers. The user-facing Adform application platform, data distribution platform, and back ends are now all running on Kubernetes. "Many APIs for critical applications are being developed for Kubernetes," says Apšega. "Teams are rewriting their applications to .NET core, because it supports containers, and preparing to move to Kubernetes. And new applications, by default, go in containers."

- -{{< case-studies/quote - image="/images/case-studies/adform/banner4.jpg" - author="Andrius Cibulskis, IT Systems Engineer, Adform" ->}} -"Releases are really nice for them, because they just push their code to Git and that's it. They don't have to worry about their virtual machines anymore." -{{< /case-studies/quote >}} - -

This big push has been driven by the real impact that these new practices have had. "Kubernetes helps our business a lot because our features are coming to market faster," says Apšega. "The deployments are very easy because developers just push the code and it automatically appears on Kubernetes." The release process went from several hours to several minutes. Autoscaling is at least six times faster than the semi-manual VM bootstrapping and application deployment required before.

- -

The team estimates that the company has experienced cost savings of 4-5x due to less hardware and fewer man hours needed to set up the hardware and virtual machines, metrics, and logging. Utilization of the hardware resources has been reduced as well, with containers notching two to three times more efficiency over virtual machines.

- -

Prometheus has also had a positive impact: "It provides high availability for metrics and alerting," says Apšega. "We monitor everything starting from hardware to applications. Having all the metrics in Grafana dashboards provides great insight on our systems."

- -{{< case-studies/quote author="Edgaras Apšega, IT Systems Engineer, Adform" >}} -"I think that our company just started our cloud native journey. It seems like a huge road ahead, but we're really happy that we joined it." -{{< /case-studies/quote >}} - -

All of these benefits have trickled down to individual team members, whose working lives have been changed for the better. "They used to have to get up at night to re-start some services, and now Kubernetes handles all of that," says Apšega. Adds Cibulskis: "Releases are really nice for them, because they just push their code to Git and that's it. They don't have to worry about their virtual machines anymore." Even the security teams have been impacted. "Security teams are always not happy," says Apšega, "and now they're happy because they can easily inspect the containers."

- -

The company plans to remain in the data centers for now, "mostly because we want to keep all the data, to not share it in any way," says Cibulskis, "and it's cheaper at our scale." But, Apšega says, the possibility of using a hybrid cloud for computing is intriguing: "One of the projects we're interested in is the Virtual Kubelet that lets you spin up the working nodes on different clouds to do some computing."

- -

Apšega, Cibulskis and their colleagues are keeping tabs on how the cloud native ecosystem develops, and are excited to contribute where they can. "I think that our company just started our cloud native journey," says Apšega. "It seems like a huge road ahead, but we're really happy that we joined it."

diff --git a/content/ko/case-studies/amadeus/amadeus_featured.png b/content/ko/case-studies/amadeus/amadeus_featured.png deleted file mode 100644 index d23d7b0163854a5da1da707c3c4aee2f43a594a5..0000000000000000000000000000000000000000 GIT binary patch literal 0 HcmV?d00001 literal 6970 zcmb_>bySqy+BczeJ%E7Z&`1w3#7H;NT@nM#016BYBMeBFgmeptbV^C5iqZ(ujdVy! z3=JPX$LGA?ALo1jIP2SM-S@rswXfgRYwf+RJ3>cGg_wYr00RSqSWQ(?@9x`s_oU%r z-Mv3~nBBcG9wL=Ykos^K(i4J!V#wLSZJ+=(XNWyi4+^pKa_fakVqjo%IU1NCO*A#Y zws2>D$ZsBgl(Xw^IEJJQ$`xY!9Et?kKXXc2X?HAWcC{R|TkpqpCLo zs_(63VC((dR>F=&Mj9ZA0^bR6h9V&Vl(UnII~XO!^0!>@-TXIMfCcb33-Y-X%fE~= z(bNGbz!6XYh+mY?76=3a#3cBEAaP+K5HCPT5C{?wln?+4@(BXLKtZsOAmGnnxzmQQ zgMsxFmH+6vTS>7vAd#+M0Rc}>Pkv8femKHj04O0LAs{FuASA?h$HC|B<${Eu_*~ps z|B;{wb+<)0x*{FnE`Z+>AvW+ANGX;(OaJA9v#X}&zXiLv{|VGx$OKRjR{QLc1P+#|Hq8~$n9?66a_UD0pGJ8Qs-aN8HoP#2_{ zq7=&=h2PH64h#}j6qJ*Y5V{+JK%kPigpi<|JV;(dOiltQCnhZS4_3{^9SL!Z=$RV^S_9{ zxBn*o-u|2T2h%|I2jjy02lHPF`tOv7KMYj6h07(We6Fs~apfZ4r-@+PLDzZ9=W-dZ zr%wzF@*`;0*D^k6G^e+2k*dsf%ufo_N1IEmqnlfQZX;lin_hb=tsFk?mhzX}cA+CO zC!SG)qBX?~(oSYxiz(YA%no(-Kx&I5b4Z(gq!z-J6~?J&E8e?)>sv2%*A(`OiIf$Q zCL??5C(YlNwV;~FgBnJq_9qF*e=UghtEuN-Op2LyD?zTjWE~7PNQrMX3-6?OdfBu;KJ{H24IoDFL!T+DKcjEF*S?sWc(M^6@IS7w5rZMuosVS1$C=`FUc7 zcB$P3JPirjJbB;~b^UUxx3Zt2f%vbL+JPV33Ez`emOK@cnH+$<+gZ?B46e*{Lg3m8 zCZwsp#g7fr{U{76wzhv~{*+~f5_;GMm`B=veSXTDd6=nkGjVZj{{Fz~nFzs=(8PL- zQ*3#rKXwoeb#E0ABy{=n?z4xg=*OqyY#-y zHg{HIz)@6#4SHkl;E8?qJ*fROyljtF%5IV#EcNnH=~n$oW|*$jh_AExGxm$V1u0{C z15|r+3;-3=QHHL1OQl(lD207V3U#=urjBiQ_ugc>^@5hxovxSUL=XANCQtQPfQi4- za%Dhj7`SZcp(mMp4D$yA(h3!hv;Mr6#ulO>%q4uy zELP3SZ}iOt>uW7P?MoY6GBY$yPYhxrg1%#SpNpIBRmrNbgnYJ>QcKzF^@@tDE8)8Z-6T<_~a>>Lr~TytJQBMk*x!{ZH6*oDH5WP7YrmWH^^FZej$ulY7{xJ_lGA%vS`n3Rb(X8Q?E$JzB$Y?X zt%8S+*36crnER3#o{kRAk!jb_9iq2;C!+>$)63b@XkY8p@V5@)TbqXt* zFw72KXs(}8L&QCn1+qD$^IeSa+A4|S){`9m(%%~{ArWBjQbl2OcM@>b9%mh7(&A07 zvJV-7`S?211}Lbw_!PL$zs1DJ`1tvan3Eu5pZH>yPRVP&qZ$zOOL?Z|x_`KwYAK=h zv?Am>(}~p~XxdcygWO}9ND%t$(pv>zJpfnG31tv*GCipcZ*HGU<1HFZtScHH@LO8! zi_D_uVlL9fg>TZW(V4Z$IBt@+L2QW62!f(5FGD$LcNSZg!k($Xm{h*z(vODKjZ?*x z^nwCIO;tziB-C%hr-W#BnAYa615L>u3p~&w4)H^`aG0R%zz3L^okE^gR&JVxM%DAf z67)EO*N{^Og`>DBVS+2{_g*A(h&RZxln1E}plnM^qV|%WyR(&{NL(?=lF7e+h$JgO z!%2Tri9$x!ucVhpGk(j9yq4pFfM$91rsh)PEM_L(kA2DRjrV;m`k#bzy4k6H{2;IO znSKJ7ilLSY(T3q1f!b}z>dI|&Ol-g7^sP5{cI_+`=G(Iv0;PCQ}F4t*1}QEd01W8RQVvXJ@3WXem9e&freC zISOm~_KWRB@4;M!OU)JF=_m7GAGZpx+Xmy+n26j5A1HX4hzMrUXamYwy)8q{Qq=ig zDs`Ky_O&Kx`o%d+t0)~Mb8o~{tMdlVKdGQfqf_@Cybt1&V|{D9l|mV|hamI}mvFg40oZU|CP2N9!)r*$|fsDMO;ZbhCGG@(TIwhkwcIPhM%^e~b zU{Gx=S=RdzuO$Cpi`45{y0Zj`fgS8Jo{33Vw!1&iirq(NtG4zaTKw2^P2mJHsKS<3 zelYn$_|`efh7snmf@VW(gpyk1dt<>IC?&Agrf6E`9h;OglS}^6F}AMAOw{ZW<}hzv zn$p-I!Pcgb-s>o}gZN_Rq=LKYM=L#$_ue&Mr3SI}%Sns6IH#${rQYDTjq<8;tr!Mt zZ0S9&`zvYGL~5;S^XTqcN5h)XBRC~A2Bc44{yYe8;C@j|^dflFg$w&3XLuS4J$_lew2rbTQoSk~wT4HC*C&H|xa@E2&4P`Q} zUM0J;lPNM6{jtS9gj^SkV#u2zKKIwQeBIIThDUsgBO7){gAGqpJU4f%TT9s48^qXF z%}=%#@z(NZhdjOVd^8S2LejWOSSNpq5NWo&t18-zs@|j?tY>;g4|s3Jjs&5hdWg(D z+0=dNOYaq(dJdjI$fmKeJzt3Tx!%$Ok2>SkE^x%b)s>A!c8w=oyTki%Yj?20O7|+1 zC(_7JBDx%X)$|}*-QSu1?7TwwCAcmBh>psD&%M4cF}dN*i->@vi9to)vT6M!M}|R% zEdf40pJVD*Jqp(2=$M<6vmX{!Pm{J++DN_CSQ&j}$h|fvXvuzHZSsYSz#?@d=#P|^ z9UJcsg-Qm7v#pA~#EO8C6Dsm%XUP@rrnJ{8%?IO?nY=4GZX=O3oJn#Fp9Re<0xPc% zs}kfl^IMEh7lhzTyY2!H8%)A(HY1)>$sCAI(*^WV``p7plE1xn^gPY>c#)3KAhoc% zNiR_PKp*D5KC$)b0A!9vG53GISV87@6Z`kr76`silF!Ypw14LV7?WL9pW@?Rdw`)* z?37qhELusbhx2-9;N~>LnY+Qyy)nDkX=EnO`pGFTUywkVY_^h|i~DLcrnIJ(Pu+L- z$yib{Y0Q0M@JO}@H+!t)x_BLP5r1q{0B>CKij9(btg;xv4qaB$Oz9YHlXOA;%Gj4pIjTWu@{p^Y~06$h!c0Uij zzliE5>@{@mf6|9$nob=1c%BO0M^y;~00OwNI%ZmEgmX|p5)mBqa{uDn%P)qZMxXhL zl_9HS%{@le^ctNHkvMt$0^NA}X}vu|*wS1Q-JoKw>DT2PZQy&^a;O044t_$){&6fP z%A12PfiPOA2(p3QZ|UXw^+gn6#0dcW@eOCZr6U2B7|{U5?X1bIgMx^X<5vwh+(7C? z+D)4Z!sWO-IwnlFEs*Uj;O+`*6Yc57{x065joaPo`tig{fT2K0jY-EbM4{9MD$-=T z!7(w|A9$dIh5$tR#=aTt_X)KOGqGJtKFK@6{aMxxE}Z#5LfDNZpYU0_Cn|fUCSzc_ zxN4Z1kE>p~y>PpRMLk}V^un5r9(xxG6cmgio{j0K?yw&boxU$G!`1ZbMfvEPG>uYz z$_!(e|N2b<&=f~k{Q8+%(sM$4K)qdL`|wx%ft5r;#1eMsRZCG=J;Tmp6e4a|o!C}3 z!RER7&x4@-J&EfcZZK;DtwkWJD3-fAK_w(~j(&9MmGOGokV}%;@k|gd z%W&cF*K+k4Y2ePR*wFc2C5ir5QSR=uO*%f0;F|kdiG>9E&#az08CGUm*)i{kY#Us> zOsH%0^Zlq)5x9fTdV0jOvC-yryGHS{h$5ELA$h(ws(Q}6E&hWLKYub*QiU=a^RhB) zMY6+kOp!%(M6*vlpMut$ zA3_z_8iaAX`W{3P0eaCJ*#0&!SbF;xPdArCoibO(JI3~!@-$bH5sI>1ByLQ=)pF8g zv|@Hg9c=DWYlMSQ|E2Q0n{fjDukO5SyAHiwIKgu4e(7k@o>2IRlCHVV6VDehY!evx z)t?_g)9Vvz-#F~}WF$VTPp0VqdUrBvtKXomeH-hzC(l-f#`bQ$je7@sXc+Y+WxaGU zHios;U7ulmPkv$)N@KV*?C%qW3n6nT3}c>;R4txzZl{l zVCo>M5!8)G2T>c6yLu7EsKLX^N>uDXm zu9bvzosHOu=*M_DTNo>`O*XbH?Jo2 z*g1n;kAjZff#!o33+?U|}2)dw7&Rhi$05OjZz99#?5${tY&Em^-jtyRq>@@&f`- zgEq>K>bkV(v|dA+9Cvz=E+&wC@j%b^dR>aGa9g@!eV=J9I|yN!obZIV@k2YnJ*%-@ z)RWMSPSYrO6jjyxVYTIDcE^VldRC(uGnJ=-=)ht0%G<1NoE}3e-!6<{Qh8}eC>H3_mg1ly~5H=#f8)JX~afSA^lwWcj1KQ&i6UXOH;?TGuahGGOedX>4LAj zKeZhgl~Gc6c-ab?ALiWTF|e-UmC1R!a2g+LsQdqDGph6+UfA8( z%_PC`n0w&hz~~2iNr=DnBsjkDA$e`OFXE6u?R%;{3G%`tlW{@_nV!{OUG>i96_!th z;ZU_1T-AXo=F5r@TUJXYZ@qisRkHoK`O(0Qw9KP9Yi`qR?o#Amfl>=S(dMUR zk5u8~M{y_1k#0(BI?-Ag4j0(f%6ZW;7X5>4@-#0j_I@-MEof(2hZQ}f`Boeil4QF@ zhNj``Wh?i9l$>lo*`UlpU)LF)I#2te<2ZUub)SjPELL^A^2xI&FREU`T9%GSUItJ( z7Trwr=IlP6X%Hih`}DE3N0&0<9!_JtQ`{G|$wXI!2CF2076tk-uQAI2m#>w3nQ{|KO%MSzD5>lF5IITgSevfKuR>jpk1%jL zPU@Rt-yx^QU-XQvCi=-khdRnNFKAwtVOr^1hM=gWrFJgoaqQ~# z{F_gBTR8!ky_IGT@_1FH&2=p79^)(@x!+W!KgF=JqXU@@m#CE{5~~w1mG;qwQj*R@ zX*tDFywQ>fCd~^?e#dQLvMuP8M%b^~OEvYJQ!@W+aUpe`q}cFE#=Q{dZDtl|O~d=_ zqqe$a;rXN9hN|vtbKgl)QqfQ0f&4mC3cT#!;v%?NIPn~^2CL5BDwS=(m^RUE7rM9K zrEgtjRcY~ls4vjz#=PK(Qcfap>V2+F*F7c9VOY%&%>y@nzp{}$xL#snv3qq%YP5K9 z6{xEnhq_-bD+)|@C43UjiHlZ`d~YtJ?`fK|Hm-d7TxD)o0g|1q{ou3wwi^1^E-s2z zVzq+#h9kznFOGt?|4b(wN+-5p+5+*40PR+eu2Y*iNUqr zQG^!Ep(>(zdk{mrKG(mODvjKg209qnzyVx7dwmH%OA}A&3`*`p^BL*u@}k}nb)k)}L6$}Rq4OD0xs zWD9bTg9MJ$xei=m^@9vaUg@Tmjz|u#&AQnf=)~T-`U9v^18*f@*UR$m7V&GPh>c92 zQAPg}idU&AGsOCS`S`KmxfS?P$ZXq^T!(8P#m=EJ^LcR&0|A;R(m6`~y|Ez-6L*C3 zI_a(vEWMZV*Bgv~HIM&4^@;zxhVy@~{H$pC_8tw!MHs%YeirHN@4s1UN?M8)^43BB E2e!!Rvj6}9 diff --git a/content/ko/case-studies/amadeus/amadeus_logo.png b/content/ko/case-studies/amadeus/amadeus_logo.png deleted file mode 100644 index 6191c7f6819f2c325bf843e85d63a35a2f4ff6ac..0000000000000000000000000000000000000000 GIT binary patch literal 0 HcmV?d00001 literal 19562 zcmeI3c|6qJ+s7wFVp3_bmPm^*V+q3y2^nO`zEoz6!7yfq8GH6pDvB&+NwTz9%UZIf zl5AzNGf^b6@3K9kx@+q0zUTKmuixwWYsM_sob$P^>wMqmI%mGVuh%!!Ku;66k#8da z003&i)eWhiZPb_l1}5q|lU4jM^@E)Nw;%!l9NcTKb%3}89soc=5@T#mGS@w>h{8Kd zB5mAtj?oeaeD`zCVPOH&QLA ztF4`)q584!>8Rge!VV-7K@kM<@bHlIkdef@+Ed*j5D-`zBrPpLRY(xMa3rLs1db^3 zEy!PS)X_wgD~3SA;Bi80agjE7Hxf)(crDS7$M@rMCj3Z*BYtN`r3msw5%xJ5%1=TLTk8F4@KlpM^D1o{hXX%BCYlO-pyPwG_V15R`hXifXQCBnj_ojK^b@zg3yRUkHWN)Yhu%u;gK(1G-2Q2De7`AQbew z+s`@w84<0HB%zhpYEl9Wp+1eJ@%wZ_#_vYU{3^1N(thqXdx`v5sl_@PH_uc9DoG|&YZL(;7*uT!@w^r(D1NDpr^y8G}*J=K-9RAD3&qMkz2c`|jNJ2}4uFb$j z7ZI(1fs2*~U7LZ6E+Se10~aj~x;6tBT|~431}<6}bZrJMx`=263|zD{=-LcibP>@C z7`SL@(6t%3=pv#OFmTb*pldU5(M3cnVBn&qLDy#BqKk-Dz`#XIgRafMMHdmRfPssa z23?zhi!LHs0RtB;4Z1c17hOcO0tPNx8gy+2F1mz(p4kt$=}xmIhs$fr~C8 zS^)zWEe*Oh0~cLHv;qb$S{iiif5o-&=S_fU9QDpW59%#`OOp>*sJ90Sq2PwP0D!kB z01yxi04%IhKL-H-H!uJ&ga7~(uLA%(@sZZgj{yMOg<9&W#-1I+30?%j=ZV$j-LNT1 zSoe8}n!>$S2&;J^QSL6?yZsCS@n_KJ!b)XZ1*6UHNtPhQgtiE?;)*(wv zJ?N;+THdb%hz?^3ou`QNQ(J|$urmP=3ClX^*k(wA;vIJnjs53+SIh4A zM)HbscbPwn)co>(gxS>oSx1XpS64je;yijFwlRYplwfJhE4P2;WJJ_(SXw9BwP5z< zE`Hw+o~cT9&9Uc9qK=jJvhdrNk%2Di((H8axnVDzZp6~L{ zn2RVU70f*Eg&zPmRPgegj{*cETE(*$Dx32zEeQ{3dD{%$D_UY7J(7wxF+tWk*9O~- zx-GYscxSJ^*)-3>a$Xc~-aX#aocCh4olfaY+iQ>qY1!K!ujr`i=&-)BGy}dKpzPAv zabxwNjbnrT>!#S^fk!F+4uU)v%qasl0s=jpe1yFUYJpy@Xqjv)W1~}jwO&PrgYij@ zzMj=J#$6#zSNVYE_HrO*wx3%sN9CkNM4-``fgfvPzc4uD{!gvv@#oL^fn# zJs;03Hs)9KY<}|a;jNjB+80+SN17HY4>`Z{o8U2&V_`$R?pXXFzf=EE83w)l*eiN0 zGt3?&+Zld(y43ld_x(p%piO}E0PBj)2+o8``-yiSRtd~au3%EvVyg~-z7I%B4ISbhYu1gPS3>fbVY8hvr=S*#O)|YL4w=#(;@gY)j z4Lv7b9ucz(wjZtBOtCfCIC>5^8qQOFoxk_82qHj+W&LsF#V6rWEWml{AHno}qsvKr z-A%8#7`#g7Mj*>^0&7tS_o%{MK?&b?o-4WDcQ{J&cE+u!v=e4{U>CM_)OLlF1V5JD&1AZ~rw zi%0~a@XMy(dG{-*d+eKTkBYXgtD=DgGGaf*|OK&`d+ju_UL)C zIt1hu!o`poPhC`1Y$y3a_q+2K-!@;$Fgp%{#Q#XGZX*>N|bwQ)*vfM74KtLqS% z@WVilOw57tu!TFhSIo9u2c7ZjMXPud7b8?RY}(_VaH06dhyq9VU|;|u@I(wD#Zi)=aEysN8m)&!oFS-GBl*UeH@a~)?8eu>z5#I>%HtJG!d!Rj!V*Stuw*vQ?4Sjq8FaSqLM zr~K-J92K^{Mqm&0!qcjqu1nplfd;x*_)Y+I(kn~@Lf&~|7i)INn>CzxyUy~E@4g54 zzR1hSC)O!4C-86TONG+~*gE1*gulARR;F}nbo`M- zhi+TqfpDdGtB_-xE~PcYKt+$XzsT3>rI!6w3OUSta+QEm)8ao`Tzk4f`p6YOg@q*Y zPP5}P%F!*ay{&v?7op;L#nSs4KGo!wn{q2}Ei2gNe{SfuP`=WWGoO=JI^_ii!_roh z5qka9MFTf6=1es*5)YV|BkPW80J|I#*^V4OYKz8iDc!r;&tS)!MS@4FO!tm#IKq@5W97)xm6(aZ zK9P-xB5%4qlzLdNA3bI{)^fP~` zb+?oc?k2xdOKeyB79Vf_#_Vhd4{>LRHWSn41PyYN&hgTx3gsb|`nQBp-lpivn7}BX z@aqypO=#AhdH=Gm*=`9zxov=kHu&{ELlyVsr9*MAh}#ZzjGPoZj(tCPD*^W6+NCop zi_<|fD&6caD%K^zMH8)dDS6hCv*aa{yq9keSzP1TIMoLph`dlvY4=^unHBC0K*yD) z=x6}aJB7%QSVy3O6i1L+varKh-HCH^`o292BH;C@-UO%W?%g+~qDX#|U)Xt=41FE! zyk7U2Ah``I?T_3~fU#K(0rd-^#S02iOapgXqu+$aR(6Xao`%3`SHzSwK3=O*J^-w| zRX`$9w2v|u9={`4R2PYJI_6X~mKdYx@gTBXU{|bxG&tQ=H_mB?@xt@lPAfk0;%uzU z1K{;hD-$0-sQ3+j%6uGoUinSFr{x4hAT9~N>&Z%QBK*+&xy~G}9c=qdcD!xfaZ{zh zVX*EZhkcZHqk;XvsT=!lM|`wh9;w9m?Pb zs2gp?`Rk1@cfXnV(zQgY4fcmJ1y|w&q@@LXKlN8OWOYsFe_0&c<4l=t}OEH-AfZtZoBTKiy}q- zJ(c?PNitb1hE*{X<7xhS!3eId#V;vJC5_G7rQ%|oy1V5XkhKLI_wmaJHD8U~*Y?HD zH}_-(-B1YTesKZDe|7G(b1<^Plze9g z=(^>nm{O4$ReW88Vt@GTORgl9i&wcUt4uv|;tO!N{=VA*?|KT2Il(MG2%YMXcY
yom;AHio%kq-(7Ju9>9%#8OT{Pn^w4n}+Rd@&yF5fbzCF8Tw#prG!Qbya!oMyE zATZNikjH`oq(?6#Z@Uf{kbs3|5-E#!!VFj1EdX%-+4C^b(|GYLp{^_$fl>W%98-E4r9+NW?T8bh zY&7TZWSB-i8;Nk(O2#dX%J@ z^wQ2sulP_IyTB)2Zp-yfc0v;CfWc$Rys-VmXAO7PkM=JPLaX|m;hyo^yw5aai={N2 z04eft+;RzPPW?;XTgciB$xPmM7Lml7i+FB{rVJ8`7#A>#?49@p8vve7dVdECJ@CwBL9^1%QU zYcr;_E~UChQ3=e#*@`)w;mlyd?D%IvEhVK^?T(?W2J?L!0-i8#ahOmUut zmHdYlnosJvJGGH1371XVLGh`N`izfu`UU?*{{3Ph zrw+^bdPW>lyJR(4R`1+c)(lKq$Ex63zod2edEk;u%))bC=wLu);~@^uk~8V^6Ng4X zvRn$UoQ|ZXjb7@!GB&up7*-;ObJnyweea}<<5F2ft5ROya{QI5&4A#P`1>0hYHa#D z&RrYxDct@V{OrMK+ru$u>+<}jJk}fi1*9oWef8ew>mvgI1S=<9Kg60>L%GhKjnwe|` z7qoy1rv-;}v9K+wO(9mjjxhsi*XlfgD~^lf^41EE9kjKYV))%Q)jndd5U7Z}n_u;z`kII&r`HZZl@fR6|MQmyns_)6dHkjO^Kd%W8T z?e;OpAKi1SaFStn@Aof-WXPQ~+iGmNd$)GM?EZ+{b)GrRV2>@YZc6VDG2HFrIT!WN zK%*`5;A~?iCz~#XyJ#-q@=`BiEOfMZ&h(-L^>nWTa?pmH8g%8f@zmnT{w-j{IP?A# zem`P`PA7zo3^*UvBM^k1%o>J{I9Ll$C7d0(TvHwUcDsrnIdcE(?AutDvi$qFX#ht9 zu)Mcesg*OUG&Jr#*<`>TbiZ_@y@k{ihyb^xmv+Zf6r*cy^C+A*r7SArE=jfCzpjPl zy&kTSS&{_f+)rlq;AXAty3HRRWC)@Zg9k44r`&-%ndQ&evRJAyu?8dHwa5MLu}awI zUjSsE0xzUja~*kFtIrdsbi00J3;yXNkhbF17+`~@CAS69S6>rr=CmsLE>~VObvmwZ zR4HZO`6hV#XOMGo>VxS7&U%2KOLtelZZup+f$wBhHIh&y2+$1fCC18SS{VZydf*O; z+eM;Ue0*#uADX-!F$X>`?Jm3ZK3QM3+kIg=vvsF-&{Nss){8o|=8nfr4SEI^<-|F* zUC$L5I~XSrwIh1KS~jH%Rz*;(#ecS-cMj%H_u7^f9-5%;Ca#HnMAY0 z{b$}RC$w8kuon(aarl*FEKU?X=9+#{@?jH?BS#%?e`P;eM_IZ2aWWKe%cDE9uuPqZ zw|UieTYBUZ#H9BiGRclpyL1G9^|Iy3Hj}(VTsJSe1b@yyuF$@(IeL9ZxnGTbp&K+W zXJGt@>&q6u@Pp^dsLSSxtcma4d$(n6&p5lOwGxI1*3vgyK3Hp-D`tJ_xb#_1bfj36 zje8dH>OpFuEDz~r&$X(uA*Oc;apMIX`{Yf|Pn8-vZoq!&3NpLGIWo+n+!8yi+Xi#- zX?tj7=mh3gL`-fiKfu25YRRU4SYZ9ilelq;lH|%t?TG*21@~h{TQvDOoA>0UzbqnX z_i7I4d0RUtH}>uJu{N=;!wrVcJrhil`dD>RD(?)UI9tY!Yww8B#B{DGS$mOJjHB+$ z`<{!u&sV|treUr1V - Another Technical Evolution for a 30-Year-Old Company -case_study_details: - - Company: Amadeus IT Group - - Location: Madrid, Spain - - Industry: Travel Technology ---- - -

Challenge

- -

In the past few years, Amadeus, which provides IT solutions to the travel industry around the world, found itself in need of a new platform for the 5,000 services supported by its service-oriented architecture. The 30-year-old company operates its own data center in Germany, and there were growing demands internally and externally for solutions that needed to be geographically dispersed. And more generally, "we had objectives of being even more highly available," says Eric Mountain, Senior Expert, Distributed Systems at Amadeus. Among the company's goals: to increase automation in managing its infrastructure, optimize the distribution of workloads, use data center resources more efficiently, and adopt new technologies more easily.

- -

Solution

- -

Mountain has been overseeing the company's migration to Kubernetes, using OpenShift Container Platform, Red Hat's enterprise container platform.

- -

Impact

- -

One of the first projects the team deployed in Kubernetes was the Amadeus Airline Cloud Availability solution, which helps manage ever-increasing flight-search volume. "It's now handling in production several thousand transactions per second, and it's deployed in multiple data centers throughout the world," says Mountain. "It's not a migration of an existing workload; it's a whole new workload that we couldn't have done otherwise. [This platform] gives us access to market opportunities that we didn't have before."

- -{{< case-studies/quote author="Eric Mountain, Senior Expert, Distributed Systems at Amadeus IT Group" >}} -"We want multi-data center capabilities, and we want them for our mainstream system as well. We didn't think that we could achieve them with our existing system. We need new automation, things that Kubernetes and OpenShift bring." -{{< /case-studies/quote >}} - -{{< case-studies/lead >}} -In his two decades at Amadeus, Eric Mountain has been the migrations guy. -{{< /case-studies/lead >}} - -

Back in the day, he worked on the company's move from Unix to Linux, and now he's overseeing the journey to cloud native. "Technology just keeps changing, and we embrace it," he says. "We are celebrating our 30 years this year, and we continue evolving and innovating to stay cost-efficient and enhance everyone's travel experience, without interrupting workflows for the customers who depend on our technology."

- -

That was the challenge that Amadeus—which provides IT solutions to the travel industry around the world, from flight searches to hotel bookings to customer feedback—faced in 2014. The technology team realized it was in need of a new platform for the 5,000 services supported by its service-oriented architecture.

- -

The tipping point occurred when they began receiving many requests, internally and externally, for solutions that needed to be geographically outside the company's main data center in Germany. "Some requests were for running our applications on customer premises," Mountain says. "There were also new services we were looking to offer that required response time to the order of a few hundred milliseconds, which we couldn't achieve with transatlantic traffic. Or at least, not without eating into a considerable portion of the time available to our applications for them to process individual queries."

- -

More generally, the company was interested in leveling up on high availability, increasing automation in managing infrastructure, optimizing the distribution of workloads and using data center resources more efficiently. "We have thousands and thousands of servers," says Mountain. "These servers are assigned roles, so even if the setup is highly automated, the machine still has a given role. It's wasteful on many levels. For instance, an application doesn't necessarily use the machine very optimally. Virtualization can help a bit, but it's not a silver bullet. If that machine breaks, you still want to repair it because it has that role and you can't simply say, 'Well, I'll bring in another machine and give it that role.' It's not fast. It's not efficient. So we wanted the next level of automation."

- -{{< case-studies/quote image="/images/case-studies/amadeus/banner3.jpg" >}} -"We hope that if we build on what others have built, what we do might actually be upstream-able. As Kubernetes and OpenShift progress, we see that we are indeed able to remove some of the additional layers we implemented to compensate for gaps we perceived earlier." -{{< /case-studies/quote >}} - -

While mainly a C++ and Java shop, Amadeus also wanted to be able to adopt new technologies more easily. Some of its developers had started using languages like Python and databases like Couchbase, but Mountain wanted still more options, he says, "in order to better adapt our technical solutions to the products we offer, and open up entirely new possibilities to our developers." Working with recent technologies and cool new things would also make it easier to attract new talent.

- -

All of those needs led Mountain and his team on a search for a new platform. "We did a set of studies and proofs of concept over a fairly short period, and we considered many technologies," he says. "In the end, we were left with three choices: build everything on premise, build on top of Kubernetes whatever happens to be missing from our point of view, or go with OpenShift and build whatever remains there."

- -

The team decided against building everything themselves—though they'd done that sort of thing in the past—because "people were already inventing things that looked good," says Mountain.

- -

Ultimately, they went with OpenShift Container Platform, Red Hat's Kubernetes-based enterprise offering, instead of building on top of Kubernetes because "there was a lot of synergy between what we wanted and the way Red Hat was anticipating going with OpenShift," says Mountain. "They were clearly developing Kubernetes, and developing certain things ahead of time in OpenShift, which were important to us, such as more security."

- -

The hope was that those particular features would eventually be built into Kubernetes, and, in the case of security, Mountain feels that has happened. "We realize that there's always a certain amount of automation that we will probably have to develop ourselves to compensate for certain gaps," says Mountain. "The less we do that, the better for us. We hope that if we build on what others have built, what we do might actually be upstream-able. As Kubernetes and OpenShift progress, we see that we are indeed able to remove some of the additional layers we implemented to compensate for gaps we perceived earlier."

- -{{< case-studies/quote image="/images/case-studies/amadeus/banner4.jpg" >}} -"It's not a migration of an existing workload; it's a whole new workload that we couldn't have done otherwise. [This platform] gives us access to market opportunities that we didn't have before." -{{< /case-studies/quote >}} - -

The first project the team tackled was one that they knew had to run outside the data center in Germany. Because of the project's needs, "We couldn't rely only on the built-in Kubernetes service discovery; we had to layer on top of that an extra service discovery level that allows us to load balance at the operation level within our system," says Mountain. They also built a stream dedicated to monitoring, which at the time wasn't offered in the Kubernetes or OpenShift ecosystem. Now that Prometheus and other products are available, Mountain says the company will likely re-evaluate their monitoring system: "We obviously always like to leverage what Kubernetes and OpenShift can offer." -

- -

The second project ended up going into production first: the Amadeus Airline Cloud Availability solution, which helps manage ever-increasing flight-search volume and was deployed in public cloud. Launched in early 2016, it is "now handling in production several thousand transactions per second, and it's deployed in multiple data centers throughout the world," says Mountain. "It's not a migration of an existing workload; it's a whole new workload that we couldn't have done otherwise. [This platform] gives us access to market opportunities that we didn't have before."

- -

Having been through this kind of technical evolution more than once, Mountain has advice on how to handle the cultural changes. "That's one aspect that we can tackle progressively," he says. "We have to go on supplying our customers with new features on our pre-existing products, and we have to keep existing products working. So we can't simply do absolutely everything from one day to the next. And we mustn't sell it that way."

- -

The first order of business, then, is to pick one or two applications to demonstrate that the technology works. Rather than choosing a high-impact, high-risk project, Mountain's team selected a smaller application that was representative of all the company's other applications in its complexity: "We just made sure we picked something that's complex enough, and we showed that it can be done."

- -{{< case-studies/quote >}} -"The bottom line is we want these multi-data center capabilities, and we want them as well for our mainstream system," he says. "And we don't think that we can implement them with our previous system. We need the new automation, homogeneity, and scale that Kubernetes and OpenShift bring." -{{< /case-studies/quote >}} - -

Next comes convincing people. "On the operations side and on the R&D side, there will be people who say quite rightly, 'There is a system, and it works, so why change?'" Mountain says. "The only thing that really convinces people is showing them the value." For Amadeus, people realized that the Airline Cloud Availability product could not have been made available on the public cloud with the company's existing system. The question then became, he says, "Do we go into a full-blown migration? Is that something that is justified?"

- -

"The bottom line is we want these multi-data center capabilities, and we want them as well for our mainstream system," he says. "And we don't think that we can implement them with our previous system. We need the new automation, homogeneity, and scale that Kubernetes and OpenShift bring."

- -

So how do you get everyone on board? "Make sure you have good links between your R&D and your operations," he says. "Also make sure you're going to talk early on to the investors and stakeholders. Figure out what it is that they will be expecting from you, that will convince them or not, that this is the right way for your company."

- -

His other advice is simply to make the technology available for people to try it. "Kubernetes and OpenShift Origin are open source software, so there's no complicated license key for the evaluation period and you're not limited to 30 days," he points out. "Just go and get it running." Along with that, he adds, "You've got to be prepared to rethink how you do things. Of course making your applications as cloud native as possible is how you'll reap the most benefits: 12 factors, CI/CD, which is continuous integration, continuous delivery, but also continuous deployment."

- -

And while they explore that aspect of the technology, Mountain and his team will likely be practicing what he preaches to others taking the cloud native journey. "See what happens when you break it, because it's important to understand the limits of the system," he says. Or rather, he notes, the advantages of it. "Breaking things on Kube is actually one of the nice things about it—it recovers. It's the only real way that you'll see that you might be able to do things."

diff --git a/content/ko/case-studies/ancestry/ancestry_featured.png b/content/ko/case-studies/ancestry/ancestry_featured.png deleted file mode 100644 index 6d63daae32139867e33f6fe0d3eb68a36f0c4ded..0000000000000000000000000000000000000000 GIT binary patch literal 0 HcmV?d00001 literal 21016 zcmeI42{hE-+rYot2uW$tVoed51!I`(TbAtm*cxNTI)<6ChNL7Zib_b75VDjkS^Fgw ziY$>`2}#z-&imEgRP+AcbN=r+?|IM6Ib-Iz&wZYI@8>?xy?5?8U!i(Bhc+Z zyvjrpju!!u6ve{DrFfCj5I90wN=)JauNVw2357{R;S!=SI0`0>g28#ezWC*s!EZ7o zJ3PujMeS>Ipd`!hK%uyypinO_FNl{ogh;Z7!lk99p)fJ1n3yP7LzL{}O2K%Gx{~*N zaq`WN3XY5=Il56CiLSgWelfO04~i^5|4KvOUth;XaQohnEBPxskRsF@;|7I8V9>u9 zX=~FqBM|;*Nv0U!evIso4U&y~+;C6>9GU1r!s5W{d;Zh~GDX!B9MV@r|H6@~C(f1f zM^|>(?|$4oNG_j2+hL(N7aRdJA%hg)e;kXO1Cc@`I}rZ@@Vol27%7hUzpz-*{M8KS z?f4h770nm3&oT97s%5}HMk$kU7z&YOL?pV%eF;~+Z@uSLR$d8DB(IP*2J7g$GNk=7 zP->$eq4UEZP6b23$*tIm!jPh3FeA7aN?ZaZE)I$)80;q}UushM&?eeB;(dPeA&!Dc zp|Xg~cL~wziTO zQ895OTvP&Mixb6&BjBPq1Oji1k+hS>Ny0x5@>lcUT&obV9xI{r)wLbyT17%!ToS2* zl$20Yfx}giiYiJ7DM?9HRcUcCaV7Y7w?CQw?iNRK^u*bzk%)xPS?-1*k#Q@@Bg_9S zd%iCVUqj(@`k@?uAF7&^Bm$05RRa5?{OA7sX8NN)pmUioYP;V@`qi`Yf9Lrp(;uF{ z(N*~|RI11Z(d&{N!G#^;@)N6-8T_AM{)3C3IstkEn^AWpgBj%WQ~N99URhoZaqhpC zep9$O{>zE^K0V-~hQeT17H&B*2Ao_O=;wv*TiK6JeX&?+=u3q^%qb3zhv9J25~5<# zIGCse7_XujJEQ~{zP5P0Bp!yrAd%nv|78C6HoP2guD_%^n5rwA*|${v(v#1%{_ccK z#8bR5B%Go>m=XVOE&Ag3SHr(IfOWvQ+T-lxpx?&!t$^A@2D&nNE{^{f=kDJvJ-r=$Diiz=VkW)#oT?F(O=W?+Y<77`pVd0QFtPWfT74a5-|2SsGF<34D?6g512CF z_i~gn(S=9?_amH~xD519#XniC?BXk{6By{2&nXQq>%Wyz*Ed^PyInj!*GG!Mz!wz0 z^7pIh&uVH@bK0gqtEo+YSQ@)JQsl(GwnkO_lhw*m;|mN!@OX`rlSClI5i(HfVrmOL zM{k^qnTjL09glC)rea2M@#b`mT&~v$8)IxgCLsI+ukR6uBpbX>IgP-*G7sDRKG z>9}a~q0-WEQ30VX(s9w|L#3tTq5?u&q~oH^he}JwMFoVmNXJE+50#dViwX#Bk&cTt zA1W;!7ZniNA{`fPK2%ydE-E0jMLI6pe5kZ^TvR}4i*#JH`A});xTt{87U{TX^P$qx zaZv%GEz)t(=0l~WA0wX&=%>qX!D`c(s5A%p)JzH z#r)&3S)42QtgILKP^?y@I3M^xEiYExKpOyj_X0rR2>_T~20sS@zyk&V?~ecg>IwjC zCq~*-ssX?@F%1<(Bk#5k9bO6?UdxPQbvWg97~NJw1yzNMd_C#&v*v^Z;XIR}0H-_; zBd3??ndn2Ed2_dn(^h+Beo9NyG-FfHS5RQkXFABL{Iqs)>AaRQYxuxm;_$%KcCYAS zzo_DAugy9a zf2p!!MqT-T+h6%wHB;fUM*2RB^H-4%MeoaP^w}+#_M*cOPu!HA$1yo}bJH_dXq(NU zbdkvHjdvT~ZRt=?!0os8-1BB^OH#R!x*3H#WDQrTzR8D9A(2y0A~S%kM;U;?n*408 zHmq*2bPeQo?&A8yTnW_Lfx*q8_S2@MMKu15o`c2=rW8v3wOVfJV@~C zzIdZ+WHhzI5r5A^!^NdPO%tfdP1s7{}r5vplM)rPYv6=-l1XysrYT zVXDARVT80_k45*xg3;?bpUJ=Qv0V<;MO-2)MKeDl}#*LIxvGor)9C zsAi2|1Y!qD5{&pS*dY*-_`LE@xSIWYH4L0`G_eb#7x0C2LNavCX#P$^ZexOQJE%X(3|Z(*Z->B7S5bRU^( zMR#umAqkHUoYNCxed!>v)uO;9@~xfy)G9y(2z-$<*5Sib>vgl<;Zyvj>p5fA@1vc| zm`pQXc3sE3Zisy^EON)}*|{(tx5my2<$#e7<_XU&3RWw!o2XqcD#tad`M8b@U%Pzy za#m($=Bhk2p})7+LT0*RHWMtnp3;?_k}}u!zRuKF=#sFfbB57VxVW>`J%{nknrLW@ zI43LV@w3*~05DW$NuJgj8D@*(o-gn09eh1_sdHZx+}LTWa+;o)$?ds^8mWyMrUz`= z-oJa>^k%T{WwX!)n2QSmae}dYn^rJkAe`V105zt!H9aVl%?B1G>T8>u^Q+q1Q_Yj= z*-`OqXP8-rOZ@$I?$eFgxej1WGm0&Csua@_agGy|lamu|H8&6MwFyZWlFi+jqT=yvt#}=Pl{dqjkU$^%u29N5i_ff)s+X#`W;`0tNEU z9#gY~9D&aqZ^~!!X$*URtz(>rv(Wghz*(3%M}A(_L`o@Ax-fTh$#}nf;}Y@NE6xfJ zr@W#MA-UaQy4R#qY|60%SuWfW#Dk7+)*ua^Y+??cYA^4qtGWw$JE^+U@7+`QHsFfkJRXm~`QcUmUN+!J{gcoe*GiRa z859ceQ|U)%eS8!Fpwa%BX6U)FQMU^T(*9z@Gk4xzDV$+If?(d?)X%U$650}zKkbn4 zemYfUnDXh8X7YWW=W)G#eZz*dm4TPUF0BcsUgy!XWB%w3$=|;jSn#p#=&N4&A z)JCg`k*-btrP!k{q69^ohqDPC_lLSoq9Io7!`%f171A^K=`PD4-qRZg%0^q8n;)k( zhfIeSPAy%%pU=83Y1hlPy!`yjBd^LXUGtjO$tgXd+;_K6#5q_AbITG^?zQNPkB^U+ z6@|fWEq7-YuLah49huIooQt{CkT+6SYg>PNZr=X_8{n4fQd)x5dT1=S7@4;(BFiL0 zD;0zG5;w zEXO}D%oiBPk`_*$Y`Up@C}Vo$pq@iv>HYhypWaj^jO^U1nCo31)cpSadsjfgwd07aVP=EsTwQ@7zT4 zFgETV$5z3Y7#OBgv=esxa|Bl8{_NSR&GP0uCHx=X-U2RSI)?yYY<9tVxvMy#T*_S| zClKn;e5`p|{^_#c6gmOgLZo!x!^M3Jk1r;?n@kn;bBnlk2TX>NY~paIWyYB>X2Xqp zxjt}oB-wK-vZKJwG(Jlbi7ePsWu@ElJtKWz94`1E@ZbfeA{)+m)*j7gZ$I#KC9)YFoE6ZBWJbPxnDgpG z{@9qD>{+D`eed6!sEHmcCPtK9F7qB2+zBWc8b&?vKFhIdJCEQN<{=So`E5J5Do+a! zT&&yfwqe}@*-t9fz$%#g9tI~I(B_)% zXw8*dKl=$Q`kL#*`&At7tamol5iA+@6-+g~k!+Ip_2JN12|> z?lg)u(HZHrf`j|vDM6e{PNWz=+u^fC7gzyHK)qDINOI)bj9z;E;VV%Iy}McffccSz zZa|FMHiv8Z0x5Ol?Hq!$5xb`%)D7F;FtRq;bQl}kM)e|4z0r4crXM_5r6rQ#)?0O= z?J zwLl>DvHlLrTZ&Cb%##f^0YDxC0b%K{%ZRM9<3bleTaLYi^-tYFJbvQraI2+kGFT;P zM}W;kMP=)Z*KSfv004n>-e42p!nYojnnZL6^1Y77pU_MTQG!5W+(!`-O$W7oz0RbO z|Cud)+*Do@_w-_(xqL@}h(abIY%SoH!z6lo<12Oh7ta?M5>o>tGnQ;M+ZL1E0oeZi zdu%rY2Vyxr_DJsE!w*mO8+v~*!mD5wRdP+_t#3x5%TctJMGOK^aG$E`;qL4bx!(Ay zOTZ!Y)5tbQ-`E)Ml*b0bmnum^SGOloBJ^-Cbv6#Y;y8_P4Db7QQO6x?ZbcUEi*6#hEX$v!Y)7vfeM~2O|PiCx5|l- z(KD8})T9v#052lb2QF)6nnd$x$r()b&XO+xJ=?d-_o!)UiAS`#;$4>>j;xDUJC>}m z(b3sCM6doGWlk%9GPbA)?d+5s5iy!{pQm;}c*_|L67b;p#SGIqy(jEpm9ARR0!466*1k3}fu~i)*e+G)+UeM-Y-xVr^t@KLp!?kq&K=1)sCZem%;x;9u~~zI zIu%DpkR_$_Yk&rZo=+dULcn7wc*GO!fr;*FPd$(FCwO|AR2Pp56fS!&S=u?6%?yvS z2_m)Qk8 ztsfd^*s)QrCUU=XuDXH6T7_zL1Jib);)L^>HyVr7&O(dv;SKxbkZy%?rdv6UMpXuTC;RAP~|Fu{QXag2IP_`KZEOlCHHB zuW2E8N4|SSk$P`jqNHSgxsUD%wWQW^IZw-Nd46vm!JUD?QMQqRwQCs%IM!gMqc^Sx zfOAKR&W*%>kSp>crEG1$q*y!S$Uf)< zw(8K-^zKs|c3uCVmn-jcV|3|C9NEV2(7;xU&D+BIhWq+56Xq6P&xIX2KaRGsFl;=BCf>CaXw^fbBh^GlJ9FpNz%1l)YOSx@KM{to@g;T@^{$Jm zy0{u}$V1}|vDo1HyzAL(ns!bRJ>?1SayHHF%9>_oUR6`A*_e@@?vkn#*SeRRbKWX7 zQ1n^HopTbEq&%Z4;Qyn-}bnH9|vf%fr{Xz^Yi}>`@ zCy`q=0lQN-9qv0B;E3}KvgUH#ZH%z-0)U*?Yr6!QV)p=nH)I8hbkLhMluraGv^F$! zU_PaQ=O@COAm;|BJmYnp9R9E^%xgO*w-hvcccem}U7SuCv^bfPvadL|JE3;)=2awz zyfr7Twr>Ie_>*q<4^oPjq!O~xuU#e`Lc1oCwFFXiv78YrD$x=g+0f{d9}ic+eS*jyBgP#<-;1a>`6| zC>pl+;!CZPGBXK}x*-DPtls7$`3pT)MszrAllD!|yYSzF)J^K!r`!M%OV(}O+Z-imm&qNSzK!n+1McdO4ZN0I0A5HC-4Zn$vD4mmXWR6j}lR^QvCL$>+E_Q*^l;V_Lz7yUG&X)l-V*u{UYw)$TdwYlJg$2W5U zfjX%NGcEJx*5%n)S!I6w5YuSfw~;4hh*E@(6xda1>yCdogH*KK>uH@chFvtuO-MLq z9$;<#qBJPT{xO#eTgV2Wr?1b-MJB8XDrRtaF7fXuk zh%6yEnajyP^rqsRaLc2i-mW7%ArU*ovlxm_LuSmF11(>kC-Xb22muN_b;11dPu%Nd ztS1sboqgw96>A0ru3v1ZzQrD2qb)>;yyP7W##mdY^!4&k^US+xhWqY2NW`+`^(X$L zoTaON#X%a|wgs7o&P<5KiN%~*WZdX}4^8MxO*mj88?bOo+04qSv-&b*`7qwKcEdi6 zyY9;cM@MbSD0eCrfo*Y}pGJxe@7nghnM&wQUB0?~FPF@V!PO@-%FLoh1_!qiN~{lT zvxlI^Fsb7DSy`^h87*04UM8h(8ySxY@3|_DCrJ!|l5a>?3}W6tsN%I*Xyv7d%`3s_ z>bhjtQa)XEEo=xIjH(!e#_?rjJ*MpXP(c}&CAFsf% zxrfdzg)G;EiKuIo(uQ12mz%REs(#|F|co~WV%aQuX=Z$n#UaKgk9~MVd*CsLv-|t|uTDx_xb5Y@8Ksj1{$n{5xMj?H zhNlqA)yX0vL$FK}3!lL^y2)h$OB6Yq2Te{06e?aZW3#C+ML^@!$;oiZtDfLbah$wO znVI|ah?N$Y=z-7i8K>BI^j~J%*a<1JFASWOzB%jTMGRHh-%-?tK5md^cmJq&lVHgr zn=&u^$t$l^TV4s>_n4}#J0l`*^NEA$z|C3+M0>;3RLSc-YR!|O?{>|XOlxjUertzM zX@|ue7=CfRBJM=c3G6LvCXL93j{~(`{U6ufEDm8*&NOk3IlQm5C|%PEjea;cV8L-! zTh{nU^$Eu5MmdX5j?cz6?Gfkb3J~0E5vgM&nUZWNYRL|6I&a1H?|DSv?Bl*v7f{$5HrL*QIZ)%ho5V+42i`~RewgTPm!uq?{!h<(^Udlz05iGFK7{wn z#j4X_2%S{oOBGprvnE(2lkhxRSop-PzJ?+)GefF{rlH09kC;VWYjAdXfWY`f`}*@e zL>H#S`@1|U-U3Hjc1Dx_K+ZqsDM}6K18~c-kKSG(-4v#0mLlss zyZQUwug=A`*`ziKv7)!KRwbN9XF^0pM5boI-)hnGvue@Ip_(60+QDZ;{ldvy@sFEH zS4;@aC4nZ4VX&MxvqkkxDZ2p5Ty`)gMLa^EFGPHAX0Wh*jSz#iuT_+0a*>VoQqJR+ z7UH-ATCm*Ks9&4UZv_`ZRR#*&f}Y?1BpJI!k4$d56(_yKbZijrF10dzW^=FWhVdt diff --git a/content/ko/case-studies/ancestry/ancestry_logo.png b/content/ko/case-studies/ancestry/ancestry_logo.png deleted file mode 100644 index 5fbade8decbc19b9e913fb9a7f0cd70d9f152ab3..0000000000000000000000000000000000000000 GIT binary patch literal 0 HcmV?d00001 literal 21141 zcmeI3c|26@-^UN88lol?S;rQYeK+>qRCbXijF}$%*^Ko6@<(&7moa>zPI>A~RsuW~&WB>qAz}1v=@V{^I ze|}(M{P$O_X9)ZcsiT@9766V??fnq}Nh!1dAkJ&8cg6XN`Xv|=W6z5~VJy(R?)Hv& zYXFdvb$3J{ZPCt93$&HBgEZ@6O#>^`8YRtYAfnE%?x={qX07Jwgx2-c&_jCKA|+6) zvNB{+?l61+d$cnG>TYl6fQ7kBvwn{Y!|T0bK33>=6K7j#R{6bz&@1X%P(_Rr8Y;po z%7YXT5`>CL@Ct}X2#O1HLk0N-MEUrI`2-|*1cYGtk1#*<=f)~ShF4NfC`*`*lFH9? z`2VC?uQ@wA!ua^y+}wEGgm^JdR(N*_2|j*7K0!eqyaf-|!@(Kh&f|b(`w`^#I7(j-HVH`z_>U|v$F0b`nCN$E_=sci5##$+2JYjxg#9;1bF%R{>cf2{N>~5 z;$-)I*HB14v>nKau{UCsxnH5zVKA#$sHYkZ5IB{GqV@)6qLyTmI2Fe~PqM^RN4kcDMc)x4oJl zZr{7Yj~0@`A1h4J35{^ZIO$<9b}~PDnbz+Jp^A!oy-I{v1j?n3Kw3NOkv%KL_gA++ z+We1*XeER*T4t|L^6*RGKY9Y9FhL2JFkXwn`1uKf{IL8>40R03+S216Vh9VugvAMC z5Lo^-27U~o5YCAIRTM&7f;cE7%o5{dk8qZ;wntc@`5Yarr1<_QB(VI|vtWuCJB$;4 zx6m>|QhfhZOyKpsx53~JSZ9O-5)D_9!58paTccn?0%%K15lcZHq@|bykFdCaIFC4j zUy#RA2#w@rX;}h8X{^?3s{a07R|LjUw{a4pNJTE&~ zJIe_E%=VXJ0$QzL;5cV zCJc9wgpkI*`T*CyhzJc1a1qkjS0CWo7ZIVs0WLxs`|1N+`ywJVIKV|nV_$uMYhOf! z1_!tZY3!>HaP5nT(BJ?UA&q_Y0j_-!5gHueBBZgeKESmvB0_@$T!b|C)d#rtMMP+D zfQyjEzWM;yzK9474sa3D*jFFm+7}U_!2vEp8vE)4T>By-G&sOTNMm1pfNNhwga!w= z2x;uA4{+^^h|u5w7a@&(^#QJZ5fK_3;3A~4uRg%FFCs#N16+hO_SFZt_C-W!aDa=D z#=iOh*S?4d4GwS-(%4rY;Mx}vp}_$zLK^$(16=zeA~ZO_MMz^`{U33W{qYhY+5!LO zpBw%aKh@bX4F2^%C{j&F9RNHz0Koqi0BnB4|IPq_3qJsSHU|J$0szopqRd)U0O0f} zTuENf{ry~uhqqo|-LtOh*`Wc!L=mb|by0AD+&lEq?K8wWX(vz6P?vZ0AJTqfQ&_Pp zHLJL7HJCc6_H^KQreC^Cx^W0Gohl?N|=q;(2aXZdPUR6MD z;nkx++C7Cg#H;_Z3$EH9VsTX9TNg9{AAE=n`715%Vo>Ne9j#{U>6&dFq*V!&I7q)> z0A;r#uxb@mm-XI;z7EBCQ;2R9VR`G>7{8>#`#SSHK8U#@KgJJ^u zKgUG(AQuD*tOTFT1z82Q{5H|qRG46+9U}?4EFn{k=&bW-G)`GtVuB zL1yBNl0$}z+S%D=#$X79NC_h6&oS4!aMf&K2X%xLo|Jn-e8^yNe!hIr-l^Io{~j?I zjxJ>m3WO-!veAjQ7Xyd`9yDKvbySI#wxh0a=g7jGaJa*>Q{C!`y1Hgn{aRGQ&0G;+ zE``pBk);Qdai{L)utC7nt3gjzoI|#@nTm^x`)tb3#lHIVi8^X0tiQ!#=9-P2Q#FHA zbrUUpa#R!b56$-SVsM<5#hV< z9wR|S#>PgId{-ibtUerXRsf!XJBaf)harnB_cI5zPmKAHbA-+bKFE7bw>sh>LSOiL zk(SyJ9>ihbh0bns~g8>Cg8L6ny5NWq zw3CFa`?uZN+AN*?7pRVu7h?)#cPjSaX!o0ZFP0h(ysJy90@h6na^t9=wq#>2XhN|htr&b z^bR$N-m;>nx<%8~ zcjfj-!{E>mcgRhy5O(D#_5pNxTzq_9PN*;~c_f=*Q1=Y^nyD8;zP`mmAlXpw`*Co%UV&2RhrL2$;tCp zuTzw@^{nC7?d?N*M^l77mZ!UsC7_MbFJZ?k?FVshgSNJ=kC&wnKk?{$Rnv;HOqKM? z&8W|K@`P3)VPH|(+Qw#UmA=95+JljecT~2E#4_SyssC+u^(ORACYU7v5F~H^s=&oS7N~H;-0{tWdoeN8mt$VdyV87`s;sOul$1

F@IXk7{R~IAl{6Vfth1*Ju5rZ-rFNgG6J<%!F0GaJMjF~sC zG9^V~mf8K)33Nae$hYN!5{O% z+;Ig~kF)8zdFZG?k%@%o%C+0QqfW5XaYt_}tgfz>S$0i|FI>olLvYLSvOZY;c=nfP zb@LoP)GD83RFBHObcSBcq1NQg!-66JKx-yM2zZhMa?}CGWp_Rn)_Fk`&P30&#G4Ax zA+3|bf`cQ~*wGU?IvV;u^%n|zIXLqGz&*`x-3?=)um9=OYr{$$Q5+gAXn4WK#?#2N zrMm2focXJm{QP|D$B)?-j2_#}luk@cpoc4XLo$`*-FHJrig4XY%6Cuk8P%fNdeFha zaEJna;7B7Zbdk4OEF#s!<1)%hUo@JZ??arPFC6{yAgsdsJ+xA z2%-EuJ)JJI`s5r8w)~O2d3wb&)W~OUx?A~foYYb@XI54pNd7lrb!}?-h;MF(8sceB zvwT`EFL$LevHMn9vD&*8RyV(95#FGy^zuQfl;rqvU>OKeA2u_y+co}1bva}O)l3I& zYzgD@mLC-HA^NDCe&;3$ub!D-^|TW1B46F|&{yMy1Q|9C4wunP4!+%;zJTm(N3NIf z>fE3g_?|xdf?4=96;=CJM3QAuYHDEMrm?B_W?Q;Ux#Q>juO9A_RAKiynz+y=C}-`^Dwu8Ts5h-6aFrb-ib}KenCAkwxGWtPfwmz=pmaAJT zr!y&3j?xK43IM7$A|a0-C#9^s^NL5!pDXa{TqrsBbmg82C4FA|%yxkGi`(=tH%afl z(0+^{RX+*9Do@Ji-gGlQqY4*)HzdTbc3^ zRceRy+4ZYmovd5`Aa;(8&1q-Ij87%osI*RgI^r7-d!V%~V$UTl{T zxe#`8E?Y@AVM_f>Zy#6071ycdk#7r(U<4O6w^XF8$8rn~*^l%X(MXdL5fE6asNknv z?sxV!8CIx&P!3E~_#zWz6k44fgECTF^2eJ`Dh4#v;IQ0S;5EJfo8^D(vnV)=%Serd{kAp&%0 zr^K?nc*HOuh9RR#Jsz))vQ3#K7X7c5w}!EhQzq>0L)yl+-@4_asmFmzHsww2eDESD9#(tMHIz zLcvxV=|Q0D6GmVNNWk-wWp)nf;hRQv$AJ6Y*Csns*V{NHNkbwoNZg<~s!2^vb22A8 zo4XFXS?h6hYrQPe?ZKi8osHK$;s(ms8Y<)8Lk&{ zr!*4rd1LIvE5q(g!bKV3a>VG^# z*dU)Gk3*p>C%;@!Wj!}g#T6WUh67CO?_knjUA_Im^Q%~F>`71LYuksrpC=4_J*HCI zDO3~_j?6J0krb|WexNcvcScPEL>cgQHdsvploF$;1_H~efp%_~ zL&e2!kEk|KOJ84n@4{EN^Lefsdpc;zT?&zAWvHK_MSUvmU5#58P*<`;ks1#G3S6xG zRej8t6fAl+mIizq!VwAj*!JqOJR)GK%V5B6Sje*TX|dH8sQ;N=oxo2Yt8+(^yp9EW zNGhN6k#BF?nxZdGZ?T=UiKXI*F@;F!ug|E=mS`B{2MAwM_d_}l?YbJ+d+g4B&`OTF z$e59MDl4H=K%mw3?Y1Is+#BA`_lz83sn$CW%Q_l@qUfpJRyPYYT^9iST5I=9b9h)A z1N}hf?WV|TSuhJxaO2^X@;i$6a!csB>Bx%$R=0Pl;YP*!7gcYcmGHPD*lpv&-Od_t zEu_?lCp9&duGn?j;?pL-!kyg2$A*T;veNhOjgFEnYsSSrsjMiaj6)AZbF^HH=MG(< zv0ab-25Dzd_AjZ|s1zo8>exNjlt~2+24}FT9-I2~y1iX84!IiIHfm3$Oq2m><;qTp%oPWL{5M`2OMlF4X=yPHzm4WNtawS~#WXG$4hDlD zwWF{%FU3jkvm_mH-sQ~_5bdY+S{hRhLYQb+!Qo33JahB&JiM^S?rl9M*v6tRwN1uZ z1}m|KOuK*R7D!AKtJT6S)y~b4YM<*H5VAVP^sZrZlgBBL7=X~wG&l@zNiNpT_NFp3 zOS%TI!z{ni9 zT|S=f;0sr38Y_C3mt!By1qK}5XF=2f5n4{C8M250=(FaR=H6Nou+*@Z)NzCJ6@&J> z#B2U{K*W^bjF-=2o_$@bkD;j5xek`I&>E#=zwxXuhsuCrQruyocC*nOP*Br;BX$IP z{`|SaDbubsxY^WV7ngE&c-zGCp`$b*nP5nSV^mWURqNNVW5=Rgy{qP*;e3Rfl}h_L zayanUeaw(K*{yfuGc#G?0op(fw*=AUwr$n@3;P9j12C1H1fs0QFfj55d&<7XKNUK_2M`?xTx~ll;00A zQ5_P^d6C+w+W&dO&ZZ`h<=2~mcd4BMxk4D{=WE=2uGeqv@cAx1)jJnTF3rLHb%G{+ zW_7Fj>*V^?cm!?OxVMa-e*ea`zVZckJ(U;LSSup`2&nA09o;|$>kEP>X-+PD*qyk@ z*U;mZ(6u=4wGNZysnqeII&(=4u3p^WZO9tA;azezIuCzqVsWr(G`W!k_`((AhvV#6 zGr4K5+TxIs@HGFMsozUe&!t$1)2oC}ZD;w^Xioz79}B%*khrK_K)q`dR?Uko)V4j> znTo&E(g~YgD2<~&$p~|Pgj>!Dyn9MaL}bfHCJw=8_S8M|(MOA`!;18x?ow&DjvW6q z!8oz`i0h72*WkQsxe;TIjsW@i`nD~v$%f4$??u%kAT@|m^012Ty-U2h=d#qIUek%y z=e7E7_Hf*UHjg@b47By;wYO`;O)hN1#Ka7;E;sCaHG9$0vU*53OVz64iE#3u5eop$ zz%MeQqB5h##uRTK^WN?Ikd~G45Ge4a7Nn*LYz7y>D+Xm}C z$DYr;~{B`oHFvFgqDql6xqnmN|mx~vB zvQ1vFcQStU3aT#=r{nhGHIzLC^0V9tdXsP&08Gj87dB0dRccfPF$p3%YflAyV=jm| zp;Yo(dS<670u`v?Eu3d?EBRJdok=a%dnV(~RJKxq4`-*WJZ|+9JjaIgTd2n${PO@M z-KeuNGL12IW9&nTLV3Eny~0NJUaH%jsmv;8Gj()y$Hu8BqbVu268iPa%FD20*>C`e zL+5FxXNb{CInXdq`uLi;8B{iR7iw`$b(now>FuHmeSxdwjD_QK+y6%IErWqVz+Vyg zNo}dhkThP%tFtT3w*oUv0?3_-&X3WKfV#7I7Zep$RejO(7|2jOV%#yAb1`0eYlDhJ z?~YD>Pfst;sk`kn z$F#x0WTKKm$_o?zP|7ID4z1qSMpn(LBX%n*_WXHWs1p8$#o_oWD468T+^t4)FXQC8 zR^BhQmw3bJPl$;-Gdg z)R9GugN>8dkW;mkj0}q)w%2VCByZwn3Iaf4UN0K$wm&}f+x|2@l{Qj}nkhv%A~*VJ zOs{!$Q5A7yq#&QLQM~18Dyy*}vlUz2>QtKvI6Y&GHn??C9(<^|4v6|VIUCHu;MN>( zoYyz|WUEmk>Bi{GzH_G|>gf3whnVBg(YiWOx;hR^jfQdR>gqK)T9Hr-@)xwXmQKkg zj#u65&cY0}QdHTCzY#j7g86NhVk@!Ck|LbG8I{rgR)1uZRGI6Lw$}C2+{iD7=Masv ztCfjfVOL9Q7uO6=MpbPlF1A`*tzUFoVkIMzBMNA6l}Nc?o25F{WhT4=UcPOG(0z+D zf2X2Lrv_e-n;H!{0urF6w8J^ta?P=VU6odb@~t1X*yLWN=>TUyMvk$NKDyDY))9C# zRll7r+<~Fu{IT1C;0)Jo0eNzTPb6xent9zGF%gqJdvq_1F~c~{;z}$-Sqke+GFsKP8s)W7T;4IUMxj3FRF|PE94a$?zM=5C3bmql_JR8(_=E*Z5 zRZDIcs8f99Cyyo9#O0nO%O~1is&m#=4~^$FechP`F?7aeLK~nB9}}&_cI(WTw~3y= zWQ8ao!l6L`^vVVKNb(ZgAdxn;-)`Ysjkp*(B(jfAgtBETXlv!kirMa@>=oAy-$g`% zHz#m5Bg?UP#FYYpM556goU!65bytV`wfM8UuQ3g8Ja;@NUX)1p+8(T|d!2MCZq1dQ zRilZ6BZ=w#BNm@eFA?>jqwPjW2hAZSH(ewWg9$e(*0>~0(Ok52yiLJ9%SAZB?}EdK sS@&}?{8NPg-cyYK!Xpp|O`H`V!SH!ew9_dc|6mA!D{Cl~D45^;A4#KXZU6uP diff --git a/content/ko/case-studies/ancestry/index.html b/content/ko/case-studies/ancestry/index.html deleted file mode 100644 index 8cab3c19c6..0000000000 --- a/content/ko/case-studies/ancestry/index.html +++ /dev/null @@ -1,92 +0,0 @@ ---- -title: Ancestry Case Study -case_study_styles: true -cid: caseStudies - -new_case_study_styles: true -heading_background: /images/case-studies/ancestry/banner1.jpg -heading_title_logo: /images/ancestry_logo.png -subheading: > - Digging Into the Past With New Technology -case_study_details: - - Company: Ancestry - - Location: Lehi, Utah - - Industry: Internet Company, Online Services ---- - -

Challenge

- -

Ancestry, the global leader in family history and consumer genomics, uses sophisticated engineering and technology to help everyone, everywhere discover the story of what led to them. The company has spent more than 30 years innovating and building products and technologies that at their core, result in real and emotional human responses. Ancestry currently serves more than 2.6 million paying subscribers, holds 20 billion historical records, 90 million family trees and more than four million people are in its AncestryDNA network, making it the largest consumer genomics DNA network in the world. The company's popular website, ancestry.com, has been working with big data long before the term was popularized. The site was built on hundreds of services, technologies and a traditional deployment methodology. "It's worked well for us in the past," says Paul MacKay, software engineer and architect at Ancestry, "but had become quite cumbersome in its processing and is time-consuming. As a primarily online service, we are constantly looking for ways to accelerate to be more agile in delivering our solutions and our products."

- -

Solution

- -

The company is transitioning to cloud native infrastructure, using Docker containerization, Kubernetes orchestration and Prometheus for cluster monitoring.

- -

Impact

- -

"Every single product, every decision we make at Ancestry, focuses on delighting our customers with intimate, sometimes life-changing discoveries about themselves and their families," says MacKay. "As the company continues to grow, the increased productivity gains from using Kubernetes has helped Ancestry make customer discoveries faster. With the move to Dockerization for example, instead of taking between 20 to 50 minutes to deploy a new piece of code, we can now deploy in under a minute for much of our code. We've truly experienced significant time savings in addition to the various features and benefits from cloud native and Kubernetes-type technologies."

- -{{< case-studies/quote author="PAUL MACKAY, SOFTWARE ENGINEER AND ARCHITECT AT ANCESTRY" >}} -"At a certain point, you have to step back if you're going to push a new technology and get key thought leaders with engineers within the organization to become your champions for new technology adoption. At training sessions, the development teams were always the ones that were saying, 'Kubernetes saved our time tremendously; it's an enabler. It really is incredible.'" -{{< /case-studies/quote >}} - -{{< case-studies/lead >}} -It started with a Shaky Leaf. -{{< /case-studies/lead >}} - -

Since its introduction a decade ago, the Shaky Leaf icon has become one of Ancestry's signature features, which signals to users that there's a helpful hint you can use to find out more about your family tree.

- -

So when the company decided to begin moving its infrastructure to cloud native technology, the first service that was launched on Kubernetes, the open source platform for managing application containers across clusters of hosts, was this hint system. Think of it as Amazon's recommended products, but instead of recommending products the company recommends records, stories, or familial connections. "It was a very important part of the site," says Ancestry software engineer and architect Paul MacKay, "but also small enough for a pilot project that we knew we could handle in a very appropriate, secure way."

- -

And when it went live smoothly in early 2016, "our deployment time for this service literally was cut down from 50 minutes to 2 or 5 minutes," MacKay adds. "The development team was just thrilled because we're focused on supplying a great experience for our customers. And that means features, it means stability, it means all those things that we need for a first-in-class type operation."

- -

The stability of that Shaky Leaf was a signal for MacKay and his team that their decision to embrace cloud native technologies was the right one for the company. With a private data center, Ancestry built its website (which launched in 1996) on hundreds of services and technologies and a traditional deployment methodology. "It worked well for us in the past, but the sum of the legacy systems became quite cumbersome in its processing and was time-consuming," says MacKay. "We were looking for other ways to accelerate, to be more agile in delivering our solutions and our products."

- -{{< case-studies/quote image="/images/case-studies/ancestry/banner3.jpg" >}} -"And when it [Kubernetes] went live smoothly in early 2016, 'our deployment time for this service literally was cut down from 50 minutes to 2 or 5 minutes,' MacKay adds. 'The development team was just thrilled because we're focused on supplying a great experience for our customers. And that means features, it means stability, it means all those things that we need for a first-in-class type operation.'" -{{< /case-studies/quote >}} - -

That need led them in 2015 to explore containerization. Ancestry engineers had already been using technology like Java and Python on Linux, so part of the decision was about making the infrastructure more Linux-friendly. They quickly decided that they wanted to go with Docker for containerization, "but it always comes down to the orchestration part of it to make it really work," says MacKay.

- -

His team looked at orchestration platforms offered by Docker Compose, Mesos and OpenStack, and even started to prototype some homegrown solutions. And then they started hearing rumblings of the imminent release of Kubernetes v1.0. "At the forefront, we were looking at the secret store, so we didn't have to manage that all ourselves, the config maps, the methodology of seamless deployment strategy," he says. "We found that how Kubernetes had done their resources, their types, their labels and just their interface was so much further advanced than the other things we had seen. It was a feature fit."

- -{{< case-studies/lead >}} -Plus, MacKay says, "I just believed in the confidence that comes with the history that Google has with containerization. So we started out right on the leading edge of it. And we haven't looked back since." -{{< /case-studies/lead >}} - -

Which is not to say that adopting a new technology hasn't come with some challenges. "Change is hard," says MacKay. "Not because the technology is hard or that the technology is not good. It's just that people like to do things like they had done [before]. You have the early adopters and you have those who are coming in later. It was a learning experience on both sides."

- -

Figuring out the best deployment operations for Ancestry was a big part of the work it took to adopt cloud native infrastructure. "We want to make sure the process is easy and also controlled in the manner that allows us the highest degree of security that we demand and our customers demand," says MacKay. "With Kubernetes and other products, there are some good solutions, but a little bit of glue is needed to bring it into corporate processes and governances. It's like having a set of gloves that are generic, but when you really do want to grab something you have to make it so it's customized to you. That's what we had to do."

- -

Their best practices include allowing their developers to deploy into development stage and production, but then controlling the aspects that need governance and auditing, such as secrets. They found that having one namespace per service is useful for achieving that containment of secrets and config maps. And for their needs, having one container per pod makes it easier to manage and to have a smaller unit of deployment. -

- -{{< case-studies/quote image="/images/case-studies/ancestry/banner4.jpg" >}} -"The success of Ancestry's first deployment of the hint system on Kubernetes helped create momentum for greater adoption of the technology." -{{< /case-studies/quote >}} - -

With that process established, the time spent on deployment was cut down to under a minute for some services. "As programmers, we have what's called REPL: read, evaluate, print, and loop, but with Kubernetes, we have CDEL: compile, deploy, execute, and loop," says MacKay. "It's a very quick loop back and a great benefit to understand that when our services are deployed in production, they're the same as what we tested in the pre-production environments. The approach of cloud native for Ancestry provides us a better ability to scale and to accommodate the business needs as work loads occur."

- -

The success of Ancestry's first deployment of the hint system on Kubernetes helped create momentum for greater adoption of the technology. "Engineers like to code, they like to do features, they don't like to sit around waiting for things to be deployed and worrying about scaling up and out and down," says MacKay. "After a while the engineers became our champions. At training sessions, the development teams were always the ones saying, 'Kubernetes saved our time tremendously; it's an enabler; it really is incredible.' Over time, we were able to convince our management that this was a transition that the industry is making and that we needed to be a part of it."

- -

A year later, Ancestry has transitioned a good number of applications to Kubernetes. "We have many different services that make up the rich environment that [the website] has from both the DNA side and the family history side," says MacKay. "We have front-end stacks, back-end stacks and back-end processing type stacks that are in the cluster."

- -

The company continues to weigh which services it will move forward to Kubernetes, which ones will be kept as is, and which will be replaced in the future and thus don't have to be moved over. MacKay estimates that the company is "approaching halfway on those features that are going forward. We don't have to do a lot of convincing anymore. It's more of an issue of timing with getting product management and engineering staff the knowledge and information that they need."

- -{{< case-studies/quote >}} -"... 'I believe in Kubernetes. I believe in containerization. I think if we can get there and establish ourselves in that world, we will be further along and far better off being agile and all the things we talk about, and it'll go forward.'" -{{< /case-studies/quote >}} - -

Looking ahead, MacKay sees Ancestry maximizing the benefits of Kubernetes in 2017. "We're very close to having everything that should be or could be in a Linux-friendly world in Kubernetes by the end of the year," he says, adding that he's looking forward to features such as federation and horizontal pod autoscaling that are currently in the works. "Kubernetes has been very wonderful for us and we continue to ride the wave."

- -

That wave, he points out, has everything to do with the vibrant Kubernetes community, which has grown by leaps and bounds since Ancestry joined it as an early adopter. "This is just a very rough way of judging it, but on Slack in June 2015, there were maybe 500 on there," MacKay says. "The last time I looked there were maybe 8,500 just on the Slack channel. There are so many major companies and different kinds of companies involved now. It's the variety of contributors, the number of contributors, the incredibly competent and friendly community."

- -

As much as he and his team at Ancestry have benefited from what he calls "the goodness and the technical abilities of many" in the community, they've also contributed information about best practices, logged bug issues and participated in the open source conversation. And they've been active in attending meetups to help educate and give back to the local tech community in Utah. Says MacKay: "We're trying to give back as far as our experience goes, rather than just code."

- -

When he meets with companies considering adopting cloud native infrastructure, the best advice he has to give from Ancestry's Kubernetes journey is this: "Start small, but with hard problems," he says. And "you need a patron who understands the vision of containerization, to help you tackle the political as well as other technical roadblocks that can occur when change is needed."

- -

With the changes that MacKay's team has led over the past year and a half, cloud native will be part of Ancestry's technological genealogy for years to come. MacKay has been such a champion of the technology that he says people have jokingly accused him of having a Kubernetes tattoo.

- -

"I really don't," he says with a laugh. "But I'm passionate. I'm not exclusive to any technology; I use whatever I need that's out there that makes us great. If it's something else, I'll use it. But right now I believe in Kubernetes. I believe in containerization. I think if we can get there and establish ourselves in that world, we will be further along and far better off being agile and all the things we talk about, and it'll go forward."

- -

He pauses. "So, yeah, I guess you can say I'm an evangelist for Kubernetes," he says. "But I'm not getting a tattoo!"

diff --git a/content/ko/case-studies/blablacar/blablacar_featured.png b/content/ko/case-studies/blablacar/blablacar_featured.png deleted file mode 100644 index cfe37257b99e73fefe25b5f6fd582feafd6dff07..0000000000000000000000000000000000000000 GIT binary patch literal 0 HcmV?d00001 literal 7867 zcmcI}WmH_-)+NCM0TLiUuoMqnedOJD-`8Vw-yYwO?lZ_OD?FB6*sT2urN0dA2%N#=RJZG?&g3n zb>(z`GyJU}1BRPBL0||7)B*5E(bNp;j1Z^4cly^B>|y_rb%6i-nC=gZ+tn1t&CA8} zXG?zrRaE}(P<#9Tpy3D&@PGULKMKRO++bjC4KN()>|}m_ah42!O2I^=oxr9DsFM~H zYWMdks#!x3P`EV|29TEib83PB78O%-h{K;G>%UkkDk6#waD=IYIapCfoc>;i3j(na zc?sl~2Yw_P)zJKKY zX$X7RePtQ26XZSELe2?l5BS%{MIis^MM~};_5RJZ_-8LX|H$ROH^cqsdjFT}{yTNw zLw}b4Y2N$bKkW~8xbJzV`{r(xaB4(DqheK*k<@aX+D?9mFxGY3815oNW+Zd& z#m*&xcvMeL*O^D>X@|;L3S~juz?0U4#BFkpH>4MNr*FfGSvwC|PjKDDfiVNQTUwq_ zhYas)ihWGI?7kw7AiPMVAXklp!?;J|T2@>&ibL4TfFJ6tJvLb+yxDv9i*7ADo*}=$<^d1aYGRXb3+!c zL2{gyWbl3kTY)-#5E0blU0YXVkD$7bV!9Ak+1&of$5@aOIm(Kq9)9-tBUW|Y`sFJ= z?tL3&EVYHPFItL2;Dtv`uM;tOyZL9Q@9aC(2DnrS7LjA}Id14=oZ}$DLXGoX)t&mL zj_e3WnKcDQH2fUpcc8#R?1-{X^BI|lvjvx#)Ry_>8YZcaAxu8ZashghjQAjd6Jz?W zS9<3Zm;&@HzLZzD$j*}56Lxx5xm>XmEZ{(SS*Il)&rbgvL~4|X`0J~qeDVCnroN!| zA%+dfUcQBcMC;io*O8^EF`&v3)8Px3_LHM6FS_FgE&KXX0r|cM?kJoDnv6KWArDfm zf{T2(L(pXVX56qxKWD-1m8m1Pg(JZJ4aIJEe)O_a`20)A_Oot+?9d`5qR=(v-T*;R zDcvxoZi!)xkx6NqB*DQM(MZ24%)=hG{%JXr+8Kas+3DuP4w(ERgx-b=VMw;WqpfWB zQ`E?4vWGROc?1~)pRQONdm1+;QuGm8wYaiIE-d-bva!_JT3jARrwmz~hcG!8H^c1d zaa&U4XQb-V{mt`eHWtJDW@nchH@Zo_v-c*2Gj)k9>IZyD5gOYQ=EfbqtaF>MIYiYO zZ1ZE#mK$-Fe4+1>U7&?*epc%9z~|vIphM!0&BcQ`GmIsaGxeIbvPuZ) z-)HKoFJbZdvLSD-X{kerik$HmZc&3uBeS~;sN%-zqZ9;4r(0mC**0S5ENkP7%dAWy zx)~pvc6c+}OfB9-^mPU689EcAnu*o~X=RtZ>oX0EUczW|dC=W9NF276;v6M1{c(NV zdZGM!Kf6q*WPF&k{?5<1s(`eW7g9Gr1G8w`lCCfbRbH5UP8AVPae{#abgyE9xX72yP%(L1eMHry#$}8_=S!#Z@Wx zX&XFmU2WQ544AT=I`@8;{f}zkWPqLzl`JOw_j5 zFJH4AI#=_%eYUOaGz&40K~bS*l1m8)*1?2Ef@1i%M?&Ie$TTT!s~)uZ#1o<(ncz3| z!#z$NVid>p)cl__R9%LrFsJJOa)MjUZrF_xr3Lm{jO$%%5z$8<7DJWR`_oKU$p(Vw zysb=cuBJ;~#D<)uZ$N?GBOD?za^? zwB2jhon)>Qb`cVPQoHwAjb3e};ufBCwcR@-1M+U`zs!Z+&{Seo2eK6Ua}$w^%KVW#q++HFoKtHhX>QTV5zu0rLEAA7VtCSwK`_=l++hfO%wA871>aqmaD>_dv| zf<7+J3rAf|mZa~_ky1&c73blqZqrH-7>-tKC=J3GO$ZIV3~F|`tuPBc9#$J$0Xy0U z+(sgVs9c<`-ZF-=Ip%<46ug2QBrNT{5^qn&R|Hp3O^;HA+rGLy`BSVp1_eP`5ql{ z*8FaTop6`c5w)Pvx+DB$9id&nU zd+g^WRMSgopO&)-HfNq5=~llTQP+gzl8qJE)^Fae3^n3}D+k81z8Vu$FB2>18+LK5 zyAp2xUeK(ev+Q3}_r5==jw!BI=7o!)u)@z*gZP$ zo~{GQ-ckg^v$Wu9!6P>KAdg}1wMeuP>n~r$xoCyi?VLBOrp0sl#HUNN!}C(l_T*72 zp{GIa=sCnS>c>Akb-lK;1S4ByV_lb+8U;lk)mqejX=q$-1tnCQg-iG;o(UJQM}7?; z%i>|d3T4gbGuLLP!^e-hJdI2yE_=Gdx(h4d;nq3Ufb?EXle^pXV&M~Epq$25$x(A& zlG=&Z!;{R)of7jtTE-`=TMovDJ zOdu%ogd6~z4|9{;Rfx}D3%;(EcOh_zSOZF-h*nX~5h6@w3uZcL42=w{&YaA8rhrk?!)Y5zNq);0$^%c>8=B399Noso9;S4aDM;tc z|0ExHd;VSoxx;x27G$$6B^O;!w(ID;6S})7jc&Clua`748YBck_n#DRul5?Fr8Y6HV@$1B{_w4*g+OB(Cy%5DR2y(TC8y0!lPhY&o zdceLV8cyx?)XY#JcRjv<2jjV+CUdI9rNP<3Y}1y1KL*R%qGm|(_&Z`5;N|dn`LAxb zN`eV1ZOe;(Iz;1pYcs<$J+;YTJ83#hc9yMSl*?f6jV@}u2%;J@8~Q`=*M@k4mUY)V zz~uF#6vpXX)#s_MHhW#WnbD!rJ(PV#rPwf>Dl5k>eq*O*pX$1q$S_fLZly@w-@yBS4l+o71gel_cw-OZHY1 zJh=w??0}~3w&r|1B{M%C?MNvL&7SLyZiR2WseD>eZd~S> zR`BFm6WEeO+a;JyJ~(K zy&TU3->w#kfBkC|GHxWaeqDRu|KhfknNY}77kz=3h6Ue}yZxaqrq>3NU60wRyH8(+(+cm?936deOJ6Oq(3v3}MR^<$j&-!MaW4MK=X9nR~(>H9KtvMkV@Y z?T3v!;)b%Ecv|Iz0SWbaw*6K}BNJ_m0MAh~(n!sO;3ACd?|n^KI&nm zj&062rS|}uy&t$t9|lH9d=pIqCS>Dt;kQn3pN9F{84C;E{*cS1xcV+0(m@{A@IGLbrC{E7}OIh45T0u7*{9E1=izl~nQOzY);kzuP%E zj_sVp#(p@1KZ?UY=tatpcY)c8G<1{rs1}i$O_2FgXoQ2<#E&gvJ-3>I&I*i|pqX~j zepLGDYnytrMTT3Mu;>UwddgBuTEGX=vXIxuf()LrC^3|e(5(C^%~nVV(lf6&Nu=>` z=;5Y;enOwf%c8Iicdzb3DLD*g!e=`xluY#Vj?SMn9e0U}hQIrpVsRU3Wum8yK=4D5 zEglCngeqj1>hW`>{(GV}1uVLHojF|@@~30 zVnkJ+=JBqnX7uvYMe{WE-COsS#l0W2`net(FR4>Rn8=$)?67EGCGGS7SKW)r#VPw=*c$y}ounIoRXs}(<)6P4aQSrk$@7b}J;!Ay5 zD0!^s+^u@X_K+7o7D_Uinpa~*!Q?RjlgBS&q~Or3UEcTl;D^(|Ery=(EiSj7` zQ}s=HVmxzIgfO>}T(4Gz~Ys1f#fcG32)l*jkM)UE~MkkC%sbiU*ePL+|76y;Uh-6n4yz0dg zV`f{L1%dWwhGI5`d|0<`i!q*Gx11`E5&v*v<%z1wu`EB5vK3e1AiPw}&Zc{^TzKYm zyFnP{mYILe+-xb{T>M)j@RRQk%J3clxpBw1_S2-f;PUUiy3?n_=WGs7TTi!*_Nd+5 zeiq0ak&QpD=|WUcB_)bwe_x;#t@BHHo-C1tR`O+;TX}#}yHSkwM2ilO@aYTcP3gQ5 zOh-xE@fPV)g~t;UWTgxB10>V5A1X$*67bZhMv`lQNhPd<}J!$V%q#g1kIZMe(>J^5;4 zjZ*J6m#@&UgG|x#%!MjmIAvcFY1p?Qu0}cz_1%pQbH_!_Zuj2#F>$!?n>||Ftubsq zQ}XZ#jOw4O8@xRQ`C2inIJd{_d`dOu;&NbfcCsEOMd@~J?A6r(*Yq+Bnut8=fBv}C ze%EH)=IkD}J&02Wsygr?KWjw0^_C?e7AyQJ6YXMj;z&^*opfv;^m&_VM0B>iCW}k6xxsj6DwXC-q{~Fw zE_FN)?c7VJUX0SP2y*|Bmo1N9K~2W=LULq$50F#F9~Y1sZ(^{V28)#@53D2J2$n0v zPqG4Yi;_z1t`7T+_roHA#=QZbj-s&~PIAHI6RW~;%(-#qcMV%60W>OJK<)sksR>qNG4%IEvot`0dm zKbpW%`>I4DUF8j_KM(8)ACmPd)^>_S(B#+c)($gZ9;<_E+q`!FuG(bw_Dm3w!@1{^ z(^X|0MoWrG?|m_IzBH058Rl^Y#jEy4_cedqcnxshNCqx~F27#H%dJu{7yO7{xMm&8 zq#w#CWqYRW)Gm9RZ!8(`8^`a-cKu#?(jWs(@o|@h7^vNn2VS z@dk=6YgD9!J3{6Qka7aO$!_NX{wAfq-k^8G343rzSi7uyE4=S}9sa&flkLu+ijl!I zcEoBX^Mi;xBN^GEaU^rrMSJIX9hrs4VrupNO-^cl% zoc9bH93dlhaUUYW)=Y^rc)_a`9aM3Q(ju9Bch_mf$hpH>HHjzc&G6<6UGM$^;rB1- z74rI`UQ;^<0Xy^;b-wE032%pz$8j%}cyC``kAN&ya83`z9jS?vDo<6XR7Y zmwk<1r zLb(eGToboWgb9|U9%UC!B{FCi`}?!WkxUh|pS1LLY2&L(_&KY{FC~V@k_56>M3;O+ zEn&thbR}>l2sP__FMJ4SzOiz6_qJ><8%eM=fSavAlKQoJ0$N`Zsl&xKoL;!uc$WC~ z*6fR3)w>U$h;$~wosXK`xS2!7bZ;{#_9r&5lct#368kato)L~eAa}atTk5aJLpo#k z&r33#1rXdVO=67&J1#aa>*0qP>kM|ii<&^t@z&1xE(T>|> zoKA!TuAq^Z2ZA8c?>JUYt<9}y(gb%!$ykDZl8dPYBN zbA-U=)@FL#-6D==UX)w`2L$!LnlW2ZhcoICuZShavm3e9B4(L}TgRg7jr^9^nk70K z%b!1DHCGCb?B&u0If=()%qjF3-H}H69ww~&Xzwfw28jM#Se#KQ{nVL~zx7yq{B9}Y z!5I7pQt1YCvs%}>@-d6~yViz*QM5%68vfQadES9zVOdsFvC2@i4Zr%0JlS3hieO#h z=qFd%bjSMu2C%egG|o0~PhW^||aj$Rkkn*{6=}9h~5tsG!}5 zM@RPXc3JojVFK#Gw9^3L^G;<;Pv1Mv$HMY<AnB} diff --git a/content/ko/case-studies/blablacar/blablacar_logo.png b/content/ko/case-studies/blablacar/blablacar_logo.png deleted file mode 100644 index 14606e036002ed9ad34469a06a8ed39f86e41d71..0000000000000000000000000000000000000000 GIT binary patch literal 0 HcmV?d00001 literal 7841 zcmd6MWl&t(wk{f6LI@JvokklP_XO7<8>eZUrlD~SF2NzVYaqC#3GNOdNaMjtf&~j6 zE;;w?bKk3a`~AMFYOOivSmXPK{+LxGT1(?40qzT26ciK!1G_l$ zTK>`D^>&6mqES#Jq`hI5U`L2My%ogP9xBOr+}6cNZ*MKhXdtZ4rw)^c*xA4Gb%W^o zYUqJ|9l;=LMrkQ}32(7S0%wT3CB3(^6BI7yEy?&VUa`mhA2EQD{$DEYj*^W3?UbRq z7QMWS8-!k%7svzV7Zjxz0rBz+gG2>{xakG>_<;aE5P)BZhmT*34Ag_y?Er1^c0s;6000II$j~YC1AE>*fHxCrf z^f!Y71P*qyhq>FkKg@ad|(iO4g~Jv;Rb#zoDI_-Q!p`kH;ARXi<_Q{i__m( z)UtDNcY)iv!06@W|71;so?YD%Y!Ce-bZ9j`*UV<*(LK(<7DT>wE8%a z@yYtWtS}a4Fm`YlXgd)_GwE>TUEXx4U9B5HY4t0qq)d$|X;& z3>DoQpY)wMXX zpaws_jp)0lA!_P0;z5L6Rf;AOdte0yi_7)A^p`WPx~f*Ol>Ku4uvmTyI(>XZTD&v* z38N>fc1@cL@~nEw#C3E}`m49qiL6Ft)>`uefs_`Zbf2K9@t#!OD}*odv!FQE?dD%V zuw0mpk+pGrrm#;mIrG{qH*$b_NGu*Nqw^Ddoi_M>V(u-uKGt9YF$rQxrMXdj^6mCK zC=E6iUd3;Y&I)x{` ztt%5~0_z{mn9fj5595E&wjAESp${V(@(lJ#1Mei_zv^Kov6j_Qq$labi$QVKKpuVw zYAls@COaax+SPwDID+m#uq+9bf7fy8dNyyIp+FR>@B~u=VJp=h?f9M` z(0HPj$`J?3YUT@PDMnXu3NPDD@lABo5!uilnrNxM&+8`*76Uo3iG$Od9pM?wb70Dk z_V3R%;ilbU?c@e}cRkl)W#}H53JUSI8z%nIl^N=G(*C7q>Ez^YCiAXukSjFA*K1)? zm5Y&V&WbX4^^NLP%gY%R*@B0*d<0DMM zhdaJdo~Q)ABxSN*mhM6s0myDC9R zflMmenlsuYA6rFB+QDdlR;c#Fb2#UZ{1)|o*EJ_4PoVZYEF?0mzk8{(r6p-n;ea?% zy`cpHD~d7P^wBAnT#dgVVUsZ7Ka8|B0lmhd*$KlUFvd}3kMxJS$w=UZmsV&CGoM!H z)*obWU@9CWVKbQNg-zTlaeK8SmIL4863gD_g9sr*v8mbx#Ouaiv+tHWr+HSR4Rgky z^@@nnM%#i;Cs(nov~cf^<|bJtrf;RaLb^Kx?h7R^yzXxL@QwFZ&DEf!K#LDqN>Zlfl!-tkZV`vs`Z~Y82I?zkf0% zcbRVSj_nPr^ig8PPTWbbE3Ppb#B6qZaFMp*eABBn=+-=f@CeP{NF!_=?|b$RivR_4 z7NdSxzk-{t(^a=f{=+qT`iddLa7ymaa{){TS!56ge(~Y#T-Z@-Q>h7=f2`G~`(j}l z$F&3UIg=)Iax|HE>*`Mh`Bxd5LaCC?zur)%-YeTuX(({63|dZnx*m~K8r)Wl@5Y@g zm$RY&^@HGwX!FC=<<0zWN4~Sqkh6F~yh(@2P%r-X0(5Ao5QKC*Y>Vg1cRZCm-xEsD zm-(Fz!ZY_}2Ew#?ly!%Q`c(0(I!{k|MY$uiL8pM{(v4{!YGgJQ;y+K}tqD?g`2?#> zB5B4L73CznR@n7KHAr5^Mep1*?NfYj-MN@^<4d0=_8N#~2|8SRK?lmvoWKq-(&IiS zO->uRS@9+x84Hfqk$(LjAo^l~ z`z?Dxfv0<{;u}8N&qrm6;;Ah;CZN249JTQAhi@T9wN7*>Dq2!$C~}Mj6w?A0gybw3 zd8_0|&DU<}E5yl_WADr+LL^XknA*vHs1CV}O!FnZPn5yydNMl4vrH)Y5ph(s8&mu) zv@4K|H^l^eWJ(j|`6c0)SR+HK2R4~Q(!Mu%(C4`F^2iNKDf3#;t;I$&z4bi8I6G8#&Xr219GlS6~{m*$1*raVjmR;@WD=jX@aDBO(D zrQ%FOJcY}rJgP3-AK;Ei#mcO4>L?d6%A`XR>8WPF`EU2Q8JxY z6wFtfw((}jG_=|3Xm~?kbF4QrcVOV|Dp5U)IfGO%GKw%<S}hX~WIZtbYwP&d9Z9 z0W3uyu65e%7c5lP!{VA;my&NjXmTs-2y^;Go_PQGON%|VeXLS-S15udI=j10p4IL> z?t%T(9AA5P=@%Q9Q4|@rqj2X{^Ck7Mq7ontqZFGHp8l^Ar$LRhD)BUkolFII?NiVo%(_g;zWisE~|J2!H(#e1&l z)$p)^?y^)fm!y0kK3i4}&LLr)Yu(2Pc_%G-%VlXMA;E&3OgD%fPMuCkHL+mBe4P5B zG&z1SDswfPdQD+Ax=9Fh0SyE_PB~|*X!Mg>KanT_>#ZtOMLb&ud_okz`3SWHpKev zO*rf%146g*v6DfswZSr2uYa^?v@HJKMW+26E8 z3ZEBw{tHSDz%r-#ZbX&~T*PVXwiuThNm9PaU^g=CjeZ)($F+i>TSxqCKuTRij_cJy-(CUfhSZ4> zR8|lV_4U)k_&`#vc)@U6eYNY*Qh14peBkATZOGUV&_!T}x}v5Oco^VOykBMi6?U(j zFWJg%AT=MO(|Pl2nbOGsC|o|=WA~T?D!SvNRdwztdU@O@>XLU3bw_7(C2?rD}Mv`}IlD51t_8g4}DI5-DF1oL)i4caeBxc9zlQkJ7ZPuAA zl$6U(kgJZ%y*m4i*h#~*dV3p)@cyTCnMplNXFc8b9~+*BgGS~IX!ZgN zMYd=Zvako_-_R3(DjN(1%fM!M9%_eG`^|Kp3`Dj;gDmIdxq^sKsYjwvHRHFJwTagj zJ|@d=k^ZPT12_i0qdwf-A*Lr}U78+xbL%rvKFR(;yrSX6tIhoAV0A~BkPH*W zW+EH)J=W~#RI<5_mr+iL%uObZOmZ-OKP@}@DOd5pE!GPDkmo0^blm4EAS{zH63qj4lcUn?jZU_R<;;Clc%vl#GuH@#rz%_$Zu^r>HiCqKFC-inMuc~FPafe194eM zv{ddX68kcMHkAu^ke7e;;!9PTN;8oCfMz~#>5DkfM&{S~@tS^g+t(LH-(6V3 zn(~fU->d@N{rH@3NPB0xu6}Baxz7qZZCN!MTChQT%}j>81U9JV|QCyPnhPQa5Pb&{GKxGzhvb#gRn{6Z6j z_7zNf;k^I%htjK&60 zokKCQeZXc?T4D8Gg@B1I{8q^f2Q8>1p}0J~g}VznY0vU|>eynBF@-`izKoksfReLvAO;rHoZO(q>%cge{QGlCghdj_T_pMUA8u$7j384peL8GsUsm zL}p-N_Q17QZ1d)`j0%COnQiLy%x-|Wsr%AwNDtDPe3N}_aZ7b>|3)W$sy46Wf}EX_ zsj5vr+KXQ;Q=E3NR$7)UGtodic&mr^oJ47r5;=s)8qFx z-hiCftms2q%Fmn(%~`#|0_wSk;9L8%KjFGYBmRj#ErqOhRFeqFA8+kba(R#@$8p-f z-LyUj{x0>(sr_}a?ve1!cI()!TdYrI;GS!s04IcJ*cTVA9*E30#GV2zO#cNCF%y`yxIJi`!`!vOT*+$v{Olu;A2@Fr) zE(H5mymjq2K8yNkWr|o5zVPp!DY=t-hE*T(d~oPp3cAaI@q-rfS4Qa&&sKD(UroRU zIgRPEA;VX0l-m*9>J#N5<;}$gfzlF!h+(A+rhJeMsKqEMoLCAH9mi?iIrl_S7Q@>O zYkFwB!9U@X$u6vLqUR?`&~pA|JB+?Y9N{&NumvBNb}i+oQBs@piG*^1ok&7^=L|N7 zooc1i=>Vjkp`9tN2`v@VlrgrQ{%=n<%S_^gJ|CBQJX8rao$Iv}_1lR@Mo4<;I4GR% z83xAQRV#}8{>KC5w#`qj`lQlf^^zXx)ZN+b678CwTWue`f zO_0t(fASg^(kxaM1zQQJziZf0-N19i-l7n<+3yJTjQtI+|Rq3Uj#|J8oqowlWyM; z_}b0!YG2hnDBE|>Yw;(B=iCZntbNONT~13~vBm-xK+d?UBta`JvBE)O#AGv@0OO`^ z!8hX_=RuCoC;q0Rz%y>F*5J9=TWfJdO8p$IN$*aIP1n%}Hp6xshDCkayUEft`<=vq zMSE!7BD=W+I7n0~!q>r#vJ8^JD&bjwn34NiU9U4i(^<#o!}iU(vs zp@lNAFFu6(wEmN1@zM{PP;%)y4{PSJ=Z6`Q1^(0RKm5a#&~jWl1F-9$!?xB*@*m9w z(*3==d$!bNw~@~&_Q8XSwn8f~ZM?gJQupP$jSV&KF*Vwyl17dGq*i>59jq~yu~GE4 zM&w#fz!giI7q&6*`wZ*gxv;UI$O?EZkPD0s$0a#q)6BwI5p}()j@vD3={;Clv|PKz z`qGd|A?1(H?oVe`$oZsQUB8v~*$J7X_;TY&;T40fCOAspH!anQ8P}?o3VnkUFaFB- zY?4-bxV>^P;E-y#3?un{kEONzHIa<^YdK`z#T3!q{bl0kLf&()?;x^qQxjA1+sJZr zZLW$S?hlz?5$xkQF9sXa%zSMordx8}ppri?q^K?T!iS1xtSsj8brXOLr(_1hMvG9m zFzV7k@5q5K&*nIFDYdPNjV6gAunE}kJmz8shcRt|o$c4e*2c9c7q=n6yNBe)7eQ`; z@Lm}0dz_qAox?W+8;dh#5w>(5Up>~>j%{l@pPt!Gkh&@Ooe?rX+&A`$ z3bMlgerf^)C?Npnc_ z8B3u`-Hk=~J2*t6V?6M!49@2}TYrGAZ4(YGle7RZ4ae!AQQ4Y<_32O_|JbnRrDES@WcmQf|% z{#X;0niFQ?A8qCwb0y^|5jZ`GO?GA%nD8R0A=#vWlMiEILEZo%Dx6`juf-4hmr*G! zw~p=Z8Y;HIJ!AqAlzrxC`lpt$2eQ@zt&J(W(Zjc)WPmeoV*yZ7s?>1pqO#W$S?yOhZ}4?>MfRxP^5#&oFY1X{ zQM&n(pdq&fLZbsTf7%>QycVo651kA?B9C#u2Qv-DEHeRsU?61&DOrpoRwp~^*x z#nrg6x1Q{$*c~%HJLSkqqLuXfDRaJ#HVJ{#IG!(ko2S!b56m&Lzn2>1428Ick~qh_ z)+)aKqybcOty$6R6LhEp$A5p*h}etE?##n0vM!H}p76KKRq-7<*F=WA`1Wn^8$#O- zk&!MRpQ(tyA9NJ$#JsJDg)4rLNFZ#@O#{;6Z}Z(fwLeQY5UGEO-wP6+V9UL|5MzOz zPgnfx{o%JV34EWVnRdkz_Rz{!hy4>$LXZJ;lX2sY+*$ac;t?3|vvNQ;Z5}g0V)!|m zeyxd?X>sQSXKzw|jPE6jKZM!)>t!6$(-2SvUTP7jQ3|13tcCEi%jxiKtq`#LpbRKrV`pa)m0SvZEQ zL!?U{;uEsR5%+z&yB)8Rv}HW7uofaJ?y5NMQaK|a2{*k${z+B{f!@l~dTF>esWdUJ z>=Z%MOCQuW!YNfa;~eM0%kzzUuZ)5aKZmcwD{jJ;hSLQ?ALW9xLLR>PTOBf06D7)V xR$j_6+^287kMH_lHOl$FCrZAyH$kXRP=F@Iv&I@2+kb|}%8D8aHF6dq{{@bPm4*NS diff --git a/content/ko/case-studies/blablacar/index.html b/content/ko/case-studies/blablacar/index.html deleted file mode 100644 index e6537a672f..0000000000 --- a/content/ko/case-studies/blablacar/index.html +++ /dev/null @@ -1,85 +0,0 @@ ---- -title: BlaBlaCar Case Study -case_study_styles: true -cid: caseStudies - -new_case_study_styles: true -heading_background: /images/case-studies/blablacar/banner1.jpg -heading_title_logo: /images/blablacar_logo.png -subheading: > - Turning to Containerization to Support Millions of Rideshares -case_study_details: - - Company: BlaBlaCar - - Location: Paris, France - - Industry: Ridesharing Company ---- - -

Challenge

- -

The world's largest long-distance carpooling community, BlaBlaCar, connects 40 million members across 22 countries. The company has been experiencing exponential growth since 2012 and needed its infrastructure to keep up. "When you're thinking about doubling the number of servers, you start thinking, 'What should I do to be more efficient?'" says Simon Lallemand, Infrastructure Engineer at BlaBlaCar. "The answer is not to hire more and more people just to deal with the servers and installation." The team knew they had to scale the platform, but wanted to stay on their own bare metal servers.

- -

Solution

- -

Opting not to shift to cloud virtualization or use a private cloud on their own servers, the BlaBlaCar team became early adopters of containerization, using the CoreOs runtime rkt, initially deployed using fleet cluster manager. Last year, the company switched to Kubernetes orchestration, and now also uses Prometheus for monitoring.

- -

Impact

- -

"Before using containers, it would take sometimes a day, sometimes two, just to create a new service," says Lallemand. "With all the tooling that we made around the containers, copying a new service now is a matter of minutes. It's really a huge gain. We are better at capacity planning in our data center because we have fewer constraints due to this abstraction between the services and the hardware we run on. For the developers, it also means they can focus only on the features that they're developing, and not on the infrastructure."

- -{{< case-studies/quote author="Simon Lallemand, Infrastructure Engineer at BlaBlaCar" >}} -"When you're switching to this cloud-native model and running everything in containers, you have to make sure that at any moment you can reboot without any downtime and without losing traffic. [With Kubernetes] our infrastructure is much more resilient and we have better availability than before." -{{< /case-studies/quote >}} - -{{< case-studies/lead >}} -For the 40 million users of BlaBlaCar, it's easy to find strangers headed in the same direction to share rides and costs. You can even choose how much "bla bla" chatter you want from a long-distance ride mate. -{{< /case-studies/lead >}} - -

Behind the scenes, though, the infrastructure was falling woefully behind the rider community's exponential growth. Founded in 2006, the company hit its current stride around 2012. "Our infrastructure was very traditional," says Infrastructure Engineer Simon Lallemand, who began working at the company in 2014. "In the beginning, it was a bit chaotic because we had to [grow] fast. But then comes the time when you have to design things to make it manageable."

- -

By 2015, the company had about 50 bare metal servers. The team was using a MySQL database and PHP, but, Lallemand says, "it was a very static way." They also utilized the configuration management system, Chef, but had little automation in its process. "When you're thinking about doubling the number of servers, you start thinking, 'What should I do to be more efficient?'" says Lallemand. "The answer is not to hire more and more people just to deal with the servers and installation."

- -

Instead, BlaBlaCar began its cloud-native journey but wasn't sure which route to take. "We could either decide to go into cloud virtualization or even use a private cloud on our own servers," says Lallemand. "But going into the cloud meant we had to make a lot of changes in our application work, and we were just not ready to make the switch from on premise to the cloud." They wanted to keep the great performance they got on bare metal, so they didn't want to go to virtualization on premise.

- -

The solution: containerization. This was early 2015 and containers were still relatively new. "It was a bold move at the time," says Lallemand. "We decided that the next servers that we would buy in the new data center would all be the same model, so we could outsource the maintenance of the servers. And we decided to go with containers and with CoreOS Container Linux as an abstraction for this hardware. It seemed future-proof to go with containers because we could see what companies were already doing with containers."

- -{{< case-studies/quote image="/images/case-studies/blablacar/banner3.jpg">}} -"With all the tooling that we made around the containers, copying a new service is a matter of minutes. It's a huge gain. For the developers, it means they can focus only on the features that they're developing and not on the infrastructure or the hour they would test their code, or the hour that it would get deployed." -{{< /case-studies/quote >}} - -

Next, they needed to choose a runtime for the containers, but "there were very few deployments in production at that time," says Lallemand. They experimented with Docker but decided to go with rkt. Lallemand explains that for BlaBlaCar, it was "much simpler to integrate things that are on rkt." At the time, the project was still pre-v1.0, so "we could speak with the developers of rkt and give them feedback. It was an advantage." Plus, he notes, rkt was very stable, even at this early stage.

- -

Once those decisions were made that summer, the company came up with a plan for implementation. First, they formed a task force to create a workflow that would be tested by three of the 10 members on Lallemand's team. But they took care to run regular workshops with all 10 members to make sure everyone was on board. "When you're focused on your product sometimes you forget if it's really user friendly, whether other people can manage to create containers too," Lallemand says. "So we did a lot of iterations to find a good workflow."

- -

After establishing the workflow, Lallemand says with a smile that "we had this strange idea that we should try the most difficult thing first. Because if it works, it will work for everything." So the first project the team decided to containerize was the database. "Nobody did that at the time, and there were really no existing tools for what we wanted to do, including building container images," he says. So the team created their own tools, such as dgr, which builds container images so that the whole team has a common framework to build on the same images with the same standards. They also revamped the service-discovery tools Nerve and Synapse; their versions, Go-Nerve and Go-Synapse, were written in Go and built to be more efficient and include new features. All of these tools were open-sourced.

- -

At the same time, the company was working to migrate its entire platform to containers with a deadline set for Christmas 2015. With all the work being done in parallel, BlaBlaCar was able to get about 80 percent of its production into containers by its deadline with live traffic running on containers during December. (It's now at 100 percent.) "It's a really busy time for traffic," says Lallemand. "We knew that by using those new servers with containers, it would help us handle the traffic."

- -

In the middle of that peak season for carpooling, everything worked well. "The biggest impact that we had was for the deployment of new services," says Lallemand. "Before using containers, we had to first deploy a new server and create configurations with Chef. It would take sometimes a day, sometimes two, just to create a new service. And with all the tooling that we made around the containers, copying a new service is a matter of minutes. So it's really a huge gain. For the developers, it means they can focus only on the features that they're developing and not on the infrastructure or the hour they would test their code, or the hour that it would get deployed."

- -{{< case-studies/quote image="/images/case-studies/blablacar/banner4.jpg" >}} -"We realized that there was a really strong community around it [Kubernetes], which meant we would not have to maintain a lot of tools of our own," says Lallemand. "It was better if we could contribute to some bigger project like Kubernetes." -{{< /case-studies/quote >}} - -

In order to meet their self-imposed deadline, one of the decisions they made was to not do any "orchestration magic" for containers in the first production alignment. Instead, they used the basic fleet tool from CoreOS to deploy their containers. (They did build a tool called GGN, which they've open-sourced, to make it more manageable for their system engineers to use.)

- -

Still, the team knew that they'd want more orchestration. "Our tool was doing a pretty good job, but at some point you want to give more autonomy to the developer team," Lallemand says. "We also realized that we don't want to be the single point of contact for developers when they want to launch new services." By the summer of 2016, they found their answer in Kubernetes, which had just begun supporting rkt implementation.

- -

After discussing their needs with their contacts at CoreOS and Google, they were convinced that Kubernetes would work for BlaBlaCar. "We realized that there was a really strong community around it, which meant we would not have to maintain a lot of tools of our own," says Lallemand. "It was better if we could contribute to some bigger project like Kubernetes." They also started using Prometheus, as they were looking for "service-oriented monitoring that could be updated nightly." Production on Kubernetes began in December 2016. "We like to do crazy stuff around Christmas," he adds with a laugh.

- -

BlaBlaCar now has about 3,000 pods, with 1200 of them running on Kubernetes. Lallemand leads a "foundations team" of 25 members who take care of the networks, databases and systems for about 100 developers. There have been some challenges getting to this point. "The rkt implementation is still not 100 percent finished," Lallemand points out. "It's really good, but there are some features still missing. We have questions about how we do things with stateful services, like databases. We know how we will be migrating some of the services; some of the others are a bit more complicated to deal with. But the Kubernetes community is making a lot of progress on that part."

- -

The team is particularly happy that they're now able to plan capacity better in the company's data center. "We have fewer constraints since we have this abstraction between the services and the hardware we run on," says Lallemand. "If we lose a server because there's a hardware problem on it, we just move the containers onto another server. It's much more efficient. We do that by just changing a line in the configuration file. And with Kubernetes, it should be automatic, so we would have nothing to do."

- -{{< case-studies/quote >}} -"If we lose a server because there's a hardware problem on it, we just move the containers onto another server. It's much more efficient. We do that by just changing a line in the configuration file. With Kubernetes, it should be automatic, so we would have nothing to do." -{{< /case-studies/quote >}} - -

And these advances ultimately trickle down to BlaBlaCar's users. "We have improved availability overall on our website," says Lallemand. "When you're switching to this cloud-native model with running everything in containers, you have to make sure that you can at any moment reboot a server or a data container without any downtime, without losing traffic. So now our infrastructure is much more resilient and we have better availability than before."

- -

Within BlaBlaCar's technology department, the cloud-native journey has created some profound changes. Lallemand thinks that the regular meetings during the conception stage and the training sessions during implementation helped. "After that everybody took part in the migration process," he says. "Then we split the organization into different 'tribes'—teams that gather developers, product managers, data analysts, all the different jobs, to work on a specific part of the product. Before, they were organized by function. The idea is to give all these tribes access to the infrastructure directly in a self-service way without having to ask. These people are really autonomous. They have responsibility of that part of the product, and they can make decisions faster."

- -

This DevOps transformation turned out to be a positive one for the company's staffers. "The team was very excited about the DevOps transformation because it was new, and we were working to make things more reliable, more future-proof," says Lallemand. "We like doing things that very few people are doing, other than the internet giants."

- -

With these changes already making an impact, BlaBlaCar is looking to split up more and more of its application into services. "I don't say microservices because they're not so micro," Lallemand says. "If we can split the responsibilities between the development teams, it would be easier to manage and more reliable, because we can easily add and remove services if one fails. You can handle it easily, instead of adding a big monolith that we still have."

- -

When Lallemand speaks to other European companies curious about what BlaBlaCar has done with its infrastructure, he tells them to come along for the ride. "I tell them that it's such a pleasure to deal with the infrastructure that we have today compared to what we had before," he says. "They just need to keep in mind their real motive, whether it's flexibility in development or reliability or so on, and then go step by step towards reaching those objectives. That's what we've done. It's important not to do technology for the sake of technology. Do it for a purpose. Our focus was on helping the developers."

diff --git a/content/ko/case-studies/blackrock/blackrock_featured.png b/content/ko/case-studies/blackrock/blackrock_featured.png deleted file mode 100644 index 3898b88c9fa43c09caea23dd985c1bab1802d986..0000000000000000000000000000000000000000 GIT binary patch literal 0 HcmV?d00001 literal 6391 zcmb_>XIN8P(>4k!N>%9~B_JZw0@4G)5PFj$y(JKe5Q0e%LKW#vM5+Q&swhpQBSl1d z2LV9==}2!81PSl83mg|HC@AQ(G~q_T zvlIBHQJ)2V^_hCjfCoEX%^Yuxal(5euqcXKju<40R|}1BMj4?Hj)Z%iC*gTu2!bl}LVOee1!xo=!Rv!|_rNLmD1rXas{pircZ-8~|6swp zDS`eoDRaHsys8*1idR}pM$`d(Q-W7kUJNWFaZ?&B!Ye5OmKB$f7MG9_l>jS9$SFvI zdH?)?fYGpyP6|eF^*_e~-jqNtc)X{AxVX2sx0tt-7zXPs4wjdf7nhI}my{F*I7D#- z4?Mz0)C0%=mj*Zr=YVzf#JgfVczboe=+Lm{r_3f=)bvfcq7z*=lhR}aV7*$l(-QJhw;KX00-y9|9dD;1yw8xfyZD? zFc|m0cJa0g29LqHU_5zMRe6Q<5Du;$zu&L_A)u$Hpyh$XBRm{XT5u&0phC>m)luQ5 zw5l9TULCHcE-eoRtI5j2q+qHy)xc5`lB$x@a%z9c!Z8kBXp{&3FImU`%BuaV>~C8@ zdjccFQCQdeC`WZH2F?4&niX9CeJ-$n)%T~Yy1wdHxVMRdVKS`a{n2hc84PGEq~f1`wJU zaGRM+Tr<82Zlsf@>HZ$}4W({eU%7u!H;x#CA`l40XZUh&6@8`(y)E$1#26j=&jd?g z{AUun1p7ClR)hUBWoc6XBk?hv|5pkH@%@M9d0r5V#fZ{wrq;T;y6H~6&y?^*b#)Wj zgxy6JdU|^1Oa-qYUjl*o@bEB&j!SOf;?dW8+Bz~Pb7gL0C69}(s(*MCOrp_fH)|Uk zop4%qBMJA<5~356ZR6v$@p*Z9G*+eC<|(PEW6B$?Y+K{9m?{6)Wn*#dQi5JfeVI>f z>hSEHVlI&G6w#Q8#(<-(HtXSX=l%>OzaIjzY+xi?V-@lSS#QR)(6sn9v&SpKE(a%P zy0HIE;d&MT;(p9kXO zX>Va+ar0f4YfVi}ZuVp%&4mj_ySuxs_b;ypUWY(_x!c=+WBzIFb2*;*hUxq_hci|i zJd9C^iNUi6z zil>EAl};knuk-T-bF#BRr)KV2qfjWx73|5yv7}cUb`Bo`=BtE;pyeHkj*sWgGt54c zVO^g(XWM#ZU6~0I^kdCpU%c@Eq^ZpFGxi)6^KPVHPPDW|fkSoyKY2>!O8QE$_UqTL zJ#vbClGx+#>x_(aFH6S{dhGlR%Wfh|gKp(mawtv}nw8FI3YO0YEPtfmd7u<!H-Br32hz&(rEM#p(poyDLmgOptrCA5GXoW!NH-5_-fG?=ZQUGy{>JlOH?A%?aNVwsVG6TR~)@=iSE`03h=&(l! z=%QsmEhucTE-o#l#w;yU8}6$IH~iHOkHJ|wM*zduqN*%jH};<7lDS^ZLtk%6Z; zt=Vq9$l4+jX8qZge(9ylOftp~tMd!YZjXH@Qp|KJhnvK!obi4cas}ck4j()7u%*ba$$p>?q0xhZQ>+5t2Ym@K1 z-xnC>xNm`Vw5ynBuAJ3ZwcVe})k-WH8XhJMS60?wZXWFnDxHCnDDJNqeD zopn>7(&ygZ-jjal34UO6cYohNg&xN}MdpH4>{3sAti(^&;$G50V#I0nR1?LSt*YFk zZYffMAU7df4^B}^B6Lqw9&yMA9PY{82|7T3tpS_T#}9Zc41K(vjG|1A7}MgC^R##V z)NSdn5tQWQ=$LD6VNokh*!q=8q|$B*Z#KMY7vd?bsqDPF*xS5w4{ub!4(X)y_TKj+ zS%}lB&VV81(ca4ddzxno-Y;J2yoMk&}dX=_Ce&)Q%LQ+_0=&$g(O5xQ& zi-4x(@`j?1^iRqWL?>Yf~7S)AP z;pJP5RO#8!`ed9aSzz6BusxTFxI>G1fBN+4A%B1Wvk=$)utzn|9zPyvY;MtT}JQT732=*7B{8D?Yh_bO-JuTiMhX@XYQOc*x$~z#D4=<}`Nd98Aqf1z%`K#JKN0=p)L;VvGx$J-45P=fs=t00Xcs|$g?n#xtkhtV z`i4k73p4XV4)koiH-oKrTG>$?r)&ws6O7#B?7I6(zIz3V6Wa>0jy9YhjwVw!uL*Js z>ZX00x=j#rle=`(?kan6Xi^S4A2~ztUY}q-KgZ5E>RnQm36bx@cNT9aq2l zZnM$qL22gm3KhT;&mywuLe@!RcGJXX&sOpWUD{TTPW2mw-u|BD#)nkn4sPpbGM-rU z5u1!PxiXt~mv@8PXGA#c0E2Qi3>E+E`2rX(iL z6?T<o+IRqGJhp$~G{Z&6a1s~@Oz8{;S<@dW)!kVV365}2tq{0h35q;wk> zH>@J(W{){)HOhqmxxlB+PA%x0yMUKZy^pnUXp4Py zbAjm4{^*WwAvZMet6{2&b916litsH+vXgsx9XKXKjP0QWH5GAt>d_U&we7*auMQnW zM5>ZQWe$PDw{M4xLM_=cOf_E@4Z1w?>_vN(3(J(1m62N_7+6=MunE-ReA|0|mrmcx znX;+8o80_X#JWa;-gKa)udn3FMi47P-GC-!!5j_mq!Lk;sYzQB@#Uvj zaogfS{W2FygJWfg*?6X+H+*^NSnjOBF&W3Kn0TU#6OGF^W{O3LEH^|~xX zQPH#JyCbhfpVq3pO!@rf%Z(JrwW$WHQ01^VQMavc&Fp8+ocX|;3q%6=hi^h0_xg!S zXpt3*@Q|vK@PZ`W_HCzrQ+pWIdk`Ii|k2XI_g%6$#4{ zw5LDIZ5-?k8?fG35!{E(sx|)-*|6MUC>HX zSqjw`ci2;r#v2ya)nU$V=23qiFe88e{+;hc8uV(D)r&1ddhycvvwD&+ zQE_>W&y4|oJK=PkAQ|+8Np(~|HAdU(1_kZ=ZuXsu>&E#)u;<1v^`o9+33oBkE5rSzugVLL--^a!!Z5wxqwq%f*M4H!0)1>2jOsKLcRziu|uq zQc}i)9&>GR-Gg!0mdYI3g;}6P<$#GkYtWz+e)%?ZuM7mbH6_gjs{3Kbcd3RPvGD}} zY^IA{S}+)F111#SY-?gNU+@2O=i(b|E9EyC(QxJs|Ho-wLkr;O?}lYW z?LMm>=4T(ej$k#cCQ`1)gERNyt&}0T8UT(?t7jm-H3LA~tnzTao^cVq^8Gb)H{&iS ze)%zy<2a1fRw8n>g#l_pjcI>$#fjmi+96fQ=y`4dwF~V<2>XdBS@MZxLWg)qtib+t zo`AgsWE7QkpcryxgOt3@M$FT!S8>x;3jDR*4x3cLJ(qIUulJ{&eon}kKQz43Xy5!$ zIY&KC67K=vl@vA(nN-8cc1UatPjG0ajmjj)9v? zfMh6IR5LTFS?w`-=(X~xb3h##eY`c(GH&pGsxG?}ZfhpcJMZOxv|deWx^lQ5MU|za zD*P%KWe{?e-tWh_^SZ}$8bpTJ}57 zt@lPK+gfa1`TRvJ<9{S#l1>``t%F4?N0nk}#E<0a?Y>h6os`e=MyLC;3_KEs-)!4#C zZizx5GZE{cr0w@6t=x%APq%s?=1+4gquBg4n6NX?sp!4Y)1u_t{G@6$+4zB^Bw|uK z!MTQ89-AAY|Lf|9-!X)pMNLppP@S|qIF}W)2>8Yj>QLgWO=6yaZd__=YL*_Om^uiu zn*bYu&r>&uaQJL`eIL3?_v9x9fd>YTi;p*W;Gw=MyfRZ_=Op$(_EBhiM7Vn^ZI%za z`Gr$a)&@o1oWhj>^0dQfu=&X4SjJ(|nWrXF=ou-e9+zUPk1jDXg>d{fYt7NP^YhAw z8p4^-pvHKy>iSgx4u>w459=A2F*7lJ4W98=w`m9Z?CtFATAzs`xPY~-O*SW z$ZQ+X^bwVAr0BD9S_xyi!cQ??%@+Wx%ne$E=N)W;1W)W~C>2b^a0gp6?8vAFvYd!R zLq7_Gc^>~rHyoka{NQ@h?;Jy8xFc`JEByc+ai|QOTX>y_ZGBB+-cGXm%F9fk^6@YX zEAP)tMbJ7lv`jMxl{Sj6+#hRh=6eNzABN*k6h|@lY;qHh7Qb1Pej4*|DRr+fyF|{n zva(9;Z0BvuZ4*fz!!Tk(SCplxN!icpOd7R7!P+=ZsWi1U1t8(Wn+Q^M;@tyKfJ>3d zEdvF=v2CqHDh-OWblL6N$W)3gKZbkNZarvTY5oFa&wlZ~+<2hAfK1yEQlTgW15f{Pm*D6 zau2=JbQku8h5y(2!7zXIZYF3UJsln05a%_v20Tf=E%2#OVBteJIi)KyIg95~x@2EE zlLuceb$B|aq>uk2ZSv=cb4AS1$BI{fz`Sm@0+9FrokzoxIvi_QSKc;TJY@%I7` z4#Mn7AEPdZ`IX2A{yI1#P~QvRiQme1|{qMW_Pry6eJ=j(g6 z09I~258j>*JVNWlsSmU8C1o-O0WPmn4jOd0_q~k~)Mvznuh!EFT z&mTs3iu}$3C=D5sTLeOt%{*|^z-5P* zEHnO?jNxX@W&hQIELE@7b81IP$;lS5gAt&VwRfU{y#yme-=Cb({pTVcP~7{+A|BBB xzZdcTuIBx>@qbs61WJ{Emn6yO$p#NjQS^0Lx5Lj>cm4h=LQ72_UIMd!_&>^Zndtxk diff --git a/content/ko/case-studies/blackrock/blackrock_logo.png b/content/ko/case-studies/blackrock/blackrock_logo.png deleted file mode 100644 index 51e914a63b259a76b0239d93c2ce872f7722e941..0000000000000000000000000000000000000000 GIT binary patch literal 0 HcmV?d00001 literal 6430 zcmc(EcT`hdvo}bWsz?zOLg-QwdXb*clps(?XE$>!3Qy4zV}Syx+nZaXv{B}k9?v>aGI)3d?6$i4 z&iV)qX(Yy8(EO4{(B0k<;Ksv~k#~1AN1{>AAPbbWt%EFhyS5PwvbB^2KNdqk5somF zjjfs|7Nz5A$96@Ax;YgVVp430N?QySXDoNDz8C zr9XrS#Q#icZ~r&7le0GJzxDk`U?*JJCoM<_<`dI$RbEKnU8}T1tznz?4Kq zMU|CR#H56TlqDpUM3i8n%0ePgVVJO(r1BqLI0orrk8*JS!)y89yej{Z_tF>a9RbO3 z6xP-iWvPP2*n|F>v$X9$=prfxlT?yY`OkR&=C%9>UH;-#ghKzLE)YN&$faffA4~rk z0^H~F_HVxh9{%=xlmp=ISir9xlwOkJ;nBsY!xeSiC%+q!X}!{8CXPqAYKh_t>C)U# zY=YC{Ba~H#m_*7&%Y}!Y-AAN!9T((3yLCX++etK(pX=O{tx@&;{d4j>I~*t&m!If! zl-4r6-FGB6?^#0bx8K;`BXKZ`k?QuET{Z5Lnj0z1%R>)>X-JK)+`f+gLV=K-CfJPd z$_r8z5KR*)A(-)n0tk1F6b}8@(B->p;eQSN_3nQl{{wYN{s#&m{~HQG{{wZ2{!irp z3(5WXp00XLzgGJnghoYGZVYCrCZiDu#1mPs{b6Atq1?)nk{B;f&xDi1!{JCqp=OFz zBgZg&%HA?INl8iX{#4n>+nqB51H}`QlQokqEm?LwuZ6H3#&gZh&FY4RQ?c)IJZ#2B zM#`O)n?Uy$W@_C`*SJ9GNtw*w32#M=70*yl1a^|a2eu8q6q&HkV1_#Ro#ufk#%onF5bFsYMlsH&2ztgDkT z7EH`p6j|T>)=svAjWMjSJG$?A@!?_C6*Lgo%pX95BOPe@&uIKLFXC|aRH@ZyYKE{lsxNUV;Gjpcw&HJ{qpbSRjn$}A^6 zQ9tW_I^5Eoscj?DUFHtTL8W#R30q^`a1u^0rd0_P6CtLqM=Qw~rK6LW#_AX0yAN`i zjqL;#Sal3ocas9;&W-yY&y0_|Q=>SudMrBx3qXLXdW?1Bwri-A5_6==Q`oINoJv$$|Y7dtVE&6Qm9jo;f}G zJ=i%qI=U7iF{>645<-R~Rxz&5fVgRyzDr9p6`CAmB{pa5x)r$X>fqp@XISY_Ph7-2 zrD@2Ld7B8=6G(OY2`O&n<45w;yH`wV{b#6KOlq7~`$R|m1@*keI|u37J7U;>-8g!* zS>iBLvz8+(f*#H1I8S0Qac5&#U2b+EHEc7Tt98S?ji_sVo=IeSur;UT3cJi>3s(?l?muFvhGbc_}uN8ot-TzE*>zF9IM34Vcqkq z`sFWVtiFWOk+rq8bzviKNi9nIp1S*Z3(RwI!P$7@3)wLjj>o^(r4P2}7~T&XYGH0H zLEj_~uQ^}L%*^B|mstDwP|HmN%d7ditgP%JOEpTv-|77PoVR155ZqOTEB<*$sjs%V z{O#LYeD0#GjJc6qA2-~#XKzzW5rW_8iI0zrh_MD9LY9_lm}87I*tG&3K-K>Q3M!9lLk$9(*i4o)UWJ z@oSRUcpYXImUEKAO0~LsxKBV}snW`w7Bs5*y!5bbjzp<_a(p}(Ot2I*Mnp}ltw*-H zTc7dqq-5I_}Boq^3@GjYOWNy75t{w2T8(Tu8e9_OEb~XL~m|wZB34B zHG^jFsjII~&6f8jOC;y1z0I@kPZjW_l2T);OW=#Qdu|!hOqSYB82UE6UrpRe+Ny`K zw$J^5fP{*}_h81YC4e@F>UOANQI5?OT_F>8%oiAIPm-*qt2f4w)LE)7z2H-v_ zcRYKJUivxTlL8XIqM#tv^Dkix2KuPEk*&!J`(!ePH)6XSs>Ds={15q=tqDp^Oe*&ZiJ^20`8KLmpS>HCKhg_Kpdb z&z~u8^z8v=v=}co*DRPcF!8-BEVjLrliBJwNS-`{^z%O6)* zxV_XFhr+ev0_?9$n4Bypn(`ldjTV`1k7^n+`JL{nK6GciaYHvfJ3}`9N>g<07f>SQ zk`mX^a(5yiro0A4$G~t`Utgc1va}Ryl_n`eLUoG~FY6j1cEX8P&E@9Jo7YxWR&d(l z7yDBfpTklv&9~#}nkjKyi0cTSG+D1v4WS7wJZi2uMQMW0x^uf&{H;y?Nwpa?$HbhtG6S0Mk&kZRjyY% zOPQ3P&rS{`baI6Kla5cfAzhSNZ$&1NOZoKX$O$eP%a#T zTdT>og|*G`#YS25Hz}5$chwOu#1^AJ<8>vx(9%d^gy9c$rg%u z2>Uk zgu30Z&CSg)tXz2ZE|4Df?#gtRE9)EpyS?TO4bw(E#JEQEur-ACB4+74qyu(g&D>t= z#I0IOXkDZZa@&0^hE8D3?-%ec`j$FqQjN_ngvRA4;zrozD!*~*T`edmcvby6CubFp zMr9&$dfMpzEw^~3p}}>0^aQv({mqpdnLzy1|Abf@$bj)1@on1~WT`}S@$vB?+c@q* zAQ0lVdmODTLDYt}{Oel9D0i zo8)x@Vh5~1aF%K>3iMiBSg4r@gEZ)WF$JWL#q*bE{4V0==9U0?YYXDB%KLgVP9}ZKp7p)U zsW9xxbE@SY#Rdh>b}WDoH=CN83dahK(PhgNKEq!`zCFV>c&&m ztUz#oT;g!YBTl$nz3cb!l7$~LwGxNFk|i)KK6@J@Yksm#!^&D(TJ^25`_G94A0@uq zm5Cij7 zl=^&skwIA-xfB}C?L=!oi({wWSE@?wnxmzmnWrW~MAP&Ts=x(>H1sxx4fOOl!M!^R zE%y`o4Yc=n+gXFkPcYT;g%)TdQgUT;p0=(FU~?|Op#sY_5u{q*tq|8J zt|q?N8HkHA1A=#AiO}hpy$<MwKLph1>vBf((J0zf_Se{p}UeRxk7W@iu*_)_(hTLpwV2A$>{6f#UN2;h&Iw7k( z%k`$&>V)7(2BEVcS{FvG=7$hQO5N&;{hhzuB2pU=5TN{Xbv`4gwXH35BT3e)!Y1)7 zNyJ)Xgjb$}1O91vw~MccC*-J;J1tnP*6+eMA_tG(*k9<=i`A}0B{f;b963Ag&c;}? z)5YA-km&pe@@KlA>`Bb;31Nd@Y`xm0U@3?dC_GQ`Jw5%^zkHR`#o&|8Lt#J4MiTjM zc+b@f{U)^^nfr>q8SYn(Jlsp>J(|Zx%JC3R2R=$a6Y2MsQ^8Ifgg+ z=q=+8L%22HGImBhB(Dp?ZgtPj`o4~ch|pHCySW6M0RhJ8*ilBlg!Vh!Vb`x;KW@A@ zKWPRO7sRnmZv>n7?(kIa;Ke}df5#5idmZ!~g~^}&*2#q8HwEEe<=l?H&nGC@(7SzC z#`U{Cki9-~S5XPfuYXzJ=raEqe~|F=#z@}G1QQdw5Abqjv?-7wQ~+5#ik9mEdG=Du zjWi;qfhRm+!X#HIDQ{Py#|nQp3UR2kN3mjrq0sqXoc8_=c8^N*cqu6j({kO&h#D2pVVXx)TJR@uw~QSoc}~%K7kcf7WZ;BPTn%bJ)fyCm7$y+#{do zqgkBtKuP1^hL(KXer@&^L#HM~Aedb$OUG4D;h%U(j{QRGQ;O!C@7uQ>8QvWo_ z^^*beXs3l5<8!t*X6K~``C^wAJ{PyTvEeDt!=tv`?W3Zi(wi)9Z*el;1Du?GXrNdw z2&?QVWoxpkCFM_B^dQgA&*zUh;F)sU92H0^D|31dUZPMnVVjf_MYe^1dPzc^kRHia zJpVov+oh!l5D+2mKR#xwit=;$DC-?}w<`V670tSs7$sz~_I z8}npkeLXjxOlg|YKYrH8Jq56T@FNm*s)WOI^wIG#@y;rZTvuIvJ;4o;pvnPih6O!p zVmQ7~u;^!bm^f`WeJEN%E@7(lEICVRVyx-I3^G(OrGx02NRn!NkBv%i%H1*Dl8iAL zVYGtoFGlyBk;kTcIPQFd8t2?faH6jnupcA$%RLSDYwj5s7*K`(FtcU-wz$~r?d=Ws zqDe9@FE8KQ#s<2KZQe|-=LMf;X>y^#q)qaSv(NSjZA-4)biOY_pFqw zgQh7_3W$a>`)|S=Z_`)LxZY%DPE;ZjHq$Llw_85FtSSt+O&{CJj@KG|^$ZOSjrTDC z3Q(`2JkbwpY2N3jN4!rKA8Ruyt}suN=f->&qqD7FneyeRYN4TOi+0?^3P&b4vQRHt!Tm2c{stPh|2C zUV9OlLqtq0On#>ee|}ebX4at0YRmfCMqW{aXB6UyeAU!zORU_wh(U1_)G%DmT&*^77!%3$v z*9U{w2UDQ?`0F&z|HAUe=gZ<587lh%T==( z!|}~U$Dt=c^D8D6Z3!{kw>T zx-W3VELSx;_q~pbp61rn~4#& z_3T@lSWQ(`IIQm)kqtSDZP?Dg5 zW!9jm%>d|PHS)5Lc%Gd4cLVXCT73T>9l`(GUZf4p1)Uilb8jm9!i}hBm;YF(D{I0F ImCSJe1$hptGynhq diff --git a/content/ko/case-studies/blackrock/index.html b/content/ko/case-studies/blackrock/index.html deleted file mode 100644 index 725d6bc057..0000000000 --- a/content/ko/case-studies/blackrock/index.html +++ /dev/null @@ -1,83 +0,0 @@ ---- -title: BlackRock Case Study -case_study_styles: true -cid: caseStudies - -new_case_study_styles: true -heading_background: /images/case-studies/blackrock/banner1.jpg -heading_title_logo: /images/blackrock_logo.png -subheading: > - Rolling Out Kubernetes in Production in 100 Days -case_study_details: - - Company: BlackRock - - Location: New York, NY - - Industry: Financial Services ---- - -

Challenge

- -

The world's largest asset manager, BlackRock operates a very controlled static deployment scheme, which has allowed for scalability over the years. But in their data science division, there was a need for more dynamic access to resources. "We want to be able to give every investor access to data science, meaning Python notebooks, or even something much more advanced, like a MapReduce engine based on Spark," says Michael Francis, a Managing Director in BlackRock's Product Group, which runs the company's investment management platform. "Managing complex Python installations on users' desktops is really hard because everyone ends up with slightly different environments. We have existing environments that do these things, but we needed to make it real, expansive and scalable. Being able to spin that up on demand, tear it down, make that much more dynamic, became a critical thought process for us. It's not so much that we had to solve our main core production problem, it's how do we extend that? How do we evolve?"

- -

Solution

- -

Drawing from what they learned during a pilot done last year using Docker environments, Francis put together a cross-sectional team of 20 to build an investor research web app using Kubernetes with the goal of getting it into production within one quarter.

- -

Impact

- -

"Our goal was: How do you give people tools rapidly without having to install them on their desktop?" says Francis. And the team hit the goal within 100 days. Francis is pleased with the results and says, "We're going to use this infrastructure for lots of other application workloads as time goes on. It's not just data science; it's this style of application that needs the dynamism. But I think we're 6-12 months away from making a [large scale] decision. We need to gain experience of running the system in production, we need to understand failure modes and how best to manage operational issues. What's interesting is that just having this technology there is changing the way our developers are starting to think about their future development."

- -{{< case-studies/quote author="Michael Francis, Managing Director, BlackRock">}} -"My message to other enterprises like us is you can actually integrate Kubernetes into an existing, well-orchestrated machinery. You don't have to throw out everything you do. And using Kubernetes made a complex problem significantly easier." -{{< /case-studies/quote >}} - -

One of the management objectives for BlackRock's Product Group employees in 2017 was to "build cool stuff." Led by Managing Director Michael Francis, a cross-sectional group of 20 did just that: They rolled out a full production Kubernetes environment and released a new investor research web app on it. In 100 days.

- -

For a company that's the world's largest asset manager, "just equipment procurement can take 100 days sometimes, let alone from inception to delivery," says Karl Wieman, a Senior System Administrator. "It was an aggressive schedule. But it moved the dial." In fact, the project achieved two goals: It solved a business problem (creating the needed web app) as well as provided real-world, in-production experience with Kubernetes, a cloud-native technology that the company was eager to explore. "It's not so much that we had to solve our main core production problem, it's how do we extend that? How do we evolve?" says Francis. The ultimate success of this project, beyond delivering the app, lies in the fact that "we've managed to integrate a radically new thought process into a controlled infrastructure that we didn't want to change."

- -

After all, in its three decades of existence, BlackRock has "a very well-established environment for managing our compute resources," says Francis. "We manage large cluster processes on machines, so we do a lot of orchestration and management for our main production processes in a way that's very cloudish in concept. We're able to manage them in a very controlled, static deployment scheme, and that has given us a huge amount of scalability."

- -

Though that works well for the core production, the company has found that some data science workloads require more dynamic access to resources. "It's a very bursty process," says Francis, who is head of data for the company's Aladdin investment management platform division.

- -

Aladdin, which connects the people, information and technology needed for money management in real time, is used internally and is also sold as a platform to other asset managers and insurance companies. "We want to be able to give every investor access to data science, meaning Python notebooks, or even something much more advanced, like a MapReduce engine based on Spark," says Francis. But "managing complex Python installations on users' desktops is really hard because everyone ends up with slightly different environments. Docker allows us to flatten that environment."

- -{{< case-studies/quote image="/images/case-studies/blackrock/banner3.jpg">}} -"We manage large cluster processes on machines, so we do a lot of orchestration and management for our main production processes in a way that's very cloudish in concept. We're able to manage them in a very controlled, static deployment scheme, and that has given us a huge amount of scalability." -{{< /case-studies/quote >}} - -

Still, challenges remain. "If you have a shared cluster, you get this storming herd problem where everyone wants to do the same thing at the same time," says Francis. "You could put limits on it, but you'd have to build an infrastructure to define limits for our processes, and the Python notebooks weren't really designed for that. We have existing environments that do these things, but we needed to make it real, expansive, and scalable. Being able to spin that up on demand, tear it down, and make that much more dynamic, became a critical thought process for us."

- -

Made up of managers from technology, infrastructure, production operations, development and information security, Francis's team was able to look at the problem holistically and come up with a solution that made sense for BlackRock. "Our initial straw man was that we were going to build everything using Ansible and run it all using some completely different distributed environment," says Francis. "That would have been absolutely the wrong thing to do. Had we gone off on our own as the dev team and developed this solution, it would have been a very different product. And it would have been very expensive. We would not have gone down the route of running under our existing orchestration system. Because we don't understand it. These guys [in operations and infrastructure] understand it. Having the multidisciplinary team allowed us to get to the right solutions and that actually meant we didn't build anywhere near the amount we thought we were going to end up building."

- -

In search of a solution in which they could manage usage on a user-by-user level, Francis's team gravitated to Red Hat's OpenShift Kubernetes offering. The company had already experimented with other cloud-native environments, but the team liked that Kubernetes was open source, and "we felt the winds were blowing in the direction of Kubernetes long term," says Francis. "Typically we make technology choices that we believe are going to be here in 5-10 years' time, in some form. And right now, in this space, Kubernetes feels like the one that's going to be there." Adds Uri Morris, Vice President of Production Operations: "When you see that the non-Google committers to Kubernetes overtook the Google committers, that's an indicator of the momentum."

- -

Once that decision was made, the major challenge was figuring out how to make Kubernetes work within BlackRock's existing framework. "It's about understanding how we can operate, manage and support a platform like this, in addition to tacking it onto our existing technology platform," says Project Manager Michael Maskallis. "All the controls we have in place, the change management process, the software development lifecycle, onboarding processes we go through—how can we do all these things?"

- -

The first (anticipated) speed bump was working around issues behind BlackRock's corporate firewalls. "One of our challenges is there are no firewalls in most open source software," says Francis. "So almost all install scripts fail in some bizarre way, and pulling down packages doesn't necessarily work." The team ran into these types of problems using Minikube and did a few small pushes back to the open source project.

- -{{< case-studies/quote image="/images/case-studies/blackrock/banner4.jpg">}} -"Typically we make technology choices that we believe are going to be here in 5-10 years' time, in some form. And right now, in this space, Kubernetes feels like the one that's going to be there." -{{< /case-studies/quote >}} - -

There were also questions about service discovery. "You can think of Aladdin as a cloud of services with APIs between them that allows us to build applications rapidly," says Francis. "It's all on a proprietary message bus, which gives us all sorts of advantages but at the same time, how does that play in a third party [platform]?"

- -

Another issue they had to navigate was that in BlackRock's existing system, the messaging protocol has different instances in the different development, test and production environments. While Kubernetes enables a more DevOps-style model, it didn't make sense for BlackRock. "I think what we are very proud of is that the ability for us to push into production is still incredibly rapid in this [new] infrastructure, but we have the control points in place, and we didn't have to disrupt everything," says Francis. "A lot of the cost of this development was thinking how best to leverage our internal tools. So it was less costly than we actually thought it was going to be."

- -

The project leveraged tools associated with the messaging bus, for example. "The way that the Kubernetes cluster will talk to our internal messaging platform is through a gateway program, and this gateway program already has built-in checks and throttles," says Morris. "We can use them to control and potentially throttle the requests coming in from Kubernetes's very elastic infrastructure to the production infrastructure. We'll continue to go in that direction. It enables us to scale as we need to from the operational perspective."

- -

The solution also had to be complementary with BlackRock's centralized operational support team structure. "The core infrastructure components of Kubernetes are hooked into our existing orchestration framework, which means that anyone in our support team has both control and visibility to the cluster using the existing operational tools," Morris explains. "That means that I don't need to hire more people."

- -

With those points established, the team created a procedure for the project: "We rolled this out first to a development environment, then moved on to a testing environment and then eventually to two production environments, in that sequential order," says Maskallis. "That drove a lot of our learning curve. We have all these moving parts, the software components on the infrastructure side, the software components with Kubernetes directly, the interconnectivity with the rest of the environment that we operate here at BlackRock, and how we connect all these pieces. If we came across issues, we fixed them, and then moved on to the different environments to replicate that until we eventually ended up in our production environment where this particular cluster is supposed to live."

- -

The team had weekly one-hour working sessions with all the members (who are located around the world) participating, and smaller breakout or deep-dive meetings focusing on specific technical details. Possible solutions would be reported back to the group and debated the following week. "I think what made it a successful experiment was people had to work to learn, and they shared their experiences with others," says Vice President and Software Developer Fouad Semaan. Then, Francis says, "We gave our engineers the space to do what they're good at. This hasn't been top-down."

- -{{< case-studies/quote >}} -"The core infrastructure components of Kubernetes are hooked into our existing orchestration framework, which means that anyone in our support team has both control and visibility to the cluster using the existing operational tools. That means that I don't need to hire more people." -{{< /case-studies/quote >}} - -

They were led by one key axiom: To stay focused and avoid scope creep. This meant that they wouldn't use features that weren't in the core of Kubernetes and Docker. But if there was a real need, they'd build the features themselves. Luckily, Francis says, "Because of the rapidity of the development, a lot of things we thought we would have to build ourselves have been rolled into the core product. [The package manager Helm is one example]. People have similar problems."

- -

By the end of the 100 days, the app was up and running for internal BlackRock users. The initial capacity of 30 users was hit within hours, and quickly increased to 150. "People were immediately all over it," says Francis. In the next phase of this project, they are planning to scale up the cluster to have more capacity.

- -

Even more importantly, they now have in-production experience with Kubernetes that they can continue to build on—and a complete framework for rolling out new applications. "We're going to use this infrastructure for lots of other application workloads as time goes on. It's not just data science; it's this style of application that needs the dynamism," says Francis. "Is it the right place to move our core production processes onto? It might be. We're not at a point where we can say yes or no, but we felt that having real production experience with something like Kubernetes at some form and scale would allow us to understand that. I think we're 6-12 months away from making a [large scale] decision. We need to gain experience of running the system in production, we need to understand failure modes and how best to manage operational issues."

- -

For other big companies considering a project like this, Francis says commitment and dedication are key: "We got the signoff from [senior management] from day one, with the commitment that we were able to get the right people. If I had to isolate what makes something complex like this succeed, I would say senior hands-on people who can actually drive it make a huge difference." With that in place, he adds, "My message to other enterprises like us is you can actually integrate Kubernetes into an existing, well-orchestrated machinery. You don't have to throw out everything you do. And using Kubernetes made a complex problem significantly easier."

diff --git a/content/ko/case-studies/buffer/buffer_featured.png b/content/ko/case-studies/buffer/buffer_featured.png deleted file mode 100644 index cd6aaba4ca6a0327e3fdc9f43385b4199ff3fccb..0000000000000000000000000000000000000000 GIT binary patch literal 0 HcmV?d00001 literal 5263 zcmcIocUTk6)<;1F5kZvR0-_)-p(aE^ganWhx`_A|M^4 z7b(&^Lg-C80#XIOWan4L-i=|#fg9Pwn372ejKh!9vUs}ca&TO$N+%VHoH5(;l; zf79C)Z{V$Mi1T*D$y*B`6+sGKa6kY7PsW102u?&dxEDg;ce!xj{dgEG0Q#MT?1&Kf z!zg2nE(qo9iU-L`$cf`5rC}hbyo98zJWNVP3?v1SlmkQL!ICoK5J@;h9uAQN{doxh z+FY$|;QDCQKXm~;gn%8HOoD^Ko}Qi(p3)M|uC`!Fd3kv-L<%e=B@R%CyWJy_v0mat zx66M?K;zwTuJ$Cdy)zMXED>wv>`q1q0G9qS1p?`Bu|&5&-2^xc?1d$PB_$xoQ~HY# zgZa;-1j66cZe)G@zwG_Dg53=7k?>%ByqmMTD-KvVo6EO?m(mWacvqY(mt90_}SYq+$ul$4y5 zs=S}v91sK&pc##K zwfDeVtGYT9K)+8MZvW4=K>tzRpS;%pYzyokdBK1&;N#`}f6M*X5D-JhuYU(G(D^(3 z@kAizU4h{4YIq(&Ma6WWj#e`C8d^&9(Y@8T-%z+@1zmEy8wV}FY^)7^{;pguEw;E- z5~G$(UUXH)m@8dO;c^(|PHNxKKj9R*%dj<*#CB{Q}{h?6HnW&!PN) zxoSQ6Ga=rukcoa<_+|Hd{Tqs_CR~fJ6z=6kfDBKx>#@;J#(;=)K_yonnF7rpJz(hf z9?$^a{wD>|H;^N}s<2jj`nYp-z-y5X6bs(U+4~~8gSe`l0y|Y|A zw$PvDh+G`V87py_itq30g$N1RF7{{bZ!N!#VlyKtWg2A(2nf{H)yZ!W4T;2+BEs<2 z%7`O|o`In-Avh?=fxdX**fR?n;A_{Ko0?*g0f!yPqt<6fyK~7x+KF~!m56gs!o$ng zYU!z^Z2a~%(se8yOE8?flgrFuxMmMV$H({&g8;2DX5iv1zMt65NyAy1h zcYZA3DCwn~Loe;a5szU4o273q68xs%H7ax+9%?#Z8Q`a{uirLd0do}w**M$SlP+sE&5HH9E@zUFX{}Xt$pinC-H@ub1ZWR^dbsE2tO}cXNNf3Kx8}xd8_p(@ zUGmN8Jwary@2=})p=mLqWXNbQ9`n~^G#Zkx5hR~~h`kGVm4}5zg?oN57f038oPYAtV%ch?o5=}eZEY=t zs)2EqA3mKn0xWA{!aT#C+pnD;FD<;^bI?KQk9HSLKP~)d>gSP*43G&DM!g3mA>uGKLXND2xH!eE0w-&)mX5NCz# zQ)JLkIDAhmM+xH3!NHMJxhHZ>2IZBOdrnQKW73%J#PjFRY3PvypxO2$O`Yeza8tTU zn}L~f(=RRk?{&OFG;c`uj;3ar6fU$u$H*c#Cf;$D%)44v;9Fwk-DXR^N!zr>ni5MN z7>%m3lCBxN1+=FIaP;4cfRG-@+=MJQaeO3B$}R1UxcLyho3b-$jbigQsetF zaqlBdjbS+txnLG0+uQy%5$-Nhr}kF3r=#;i(@$}lfbr`4KrrqDn?&z?Jcgyk9P;i* z+ROUNTwG(3L!XH5e5A^o2lQ-jgmvq$KUx#uY6=OVnckWR<@2!oX=KL7k-6RG@z98? zxIFnMtWL+XX=6lYdof1b)F#tv=CmR`$t(z^^(o7=`0CZGpU>20*NSgicU7v35 z6$0&FasOb2d~V_aq)^9!?BQbPWFQ|3CHpF;!_03uVp$kKA`^Y_ass<-o zODo#D7@>Z<&004#G#q-;Fsq{#)Rlo|Ed7H+Cr1O04(5B(4Fb{^W3|mvmlKsHG6H!_ zPijMyCepOx^!09|$}kCXcsyQG^dyY?R^e^TF!A$z8W*tyJ67p;?bS<`mtRrnFh*KXjvnR(pU!(T7S3@mUKDDeV9BlR&M#b`07xN+gDWk-L@fya4 zQ5X4Ng1m7ksp4sVq({{kw1Kx1Qx}Dw)m(!PI_h;fEjp(A#%R0m)s-nfaf_b0Cv4z| zurRCkeTYzbQa5eADkf~b=gGiEfE1Pu0xh7KqfMp^X&x}0wvJFBFaOdrte%iFfpOQ{|>cETk*NW!8 z-7e}>)&0H!Nb-$KD#LmC`8?ijv69r5rh4mk7C`~;`BhKF>l2Vq!z?KRo6l6AzZ>i~ zwz09XvlE^|XHQ!s%`qFfOgBBBVbwP?3-_t!zw;5jMtAFA?`K^|l>RMZppaH<^&(}2 zq3*8uR?KsfH$}k9+Z#^J%@mETDZZ0$qisYh&P{xcFk5fZIWjn%S%l|r;M%w`F)_h= zc#z+r=q4t2F?f_$(YM4CqpYm_C>RocuDbPbqAt|MP&fSz`MQ{x(@=rE*BEE6hsN+) z@YaP#SuJk|brgP}f%R5DObW?(g9Xu95ho6E@kP7Av){G_Q&Uq{oPLGJ2o4oGLLGWD zTn)5oW!z@9E(WOGA^a4DRO}uJeZI&s(p(i8?~@aAkD}vCsy$hOrpd5t7$S!~ed?|e zSOV192;2}QB@?I>Nr{QvH>`GB#5W&X^EN9vO>oqr=4v4)nYzwI{#tJVNB8-J>M@V+ z?CdB7pT=8TciMR0C#KR!i~#5ZC|q#}ACI@HPeszXGafLahEkM*V`JGPU!2k(>e1@| zz_GFi8B;vm!P&NV-i9%al@%MGRC|5z8SPZ=vkB6ZK-}!>w2v*OR=&%H1C=|IPfndW zl~b`b^x1uJ>8-}_{n^%_Q}in(QxTH5hP!OvBY>iDQPp#W()=sGp%YTVe(6$3IU_f$ z^C*(gnzVD5E+&R!iKy)F@8342tYj%!c)mD8Zq#M-dl&DuPo}hlJ7d$&hGTynBLeu- zuPTohnl8SFx`zD-fAZuqXVi;9)@O9g4{c_?z8FNo(7ow}v+PdMp@tEyxjx$~eMO$C z!Y*oBB9<#o+4eAxrPxLZ2HU3l+Ve-0$ZBz4+nj+P z(X#4QZ);_>q>F0Cv$2c}Vf`OB2Y3A&C01vCdJJfjHZ{77EJS#V}N3CIzi!}Jdu`)`Yy zbY<{U@-*Psh^qyJVL5*Onmgf$q$azuK*O$frNGh$0p4Zjyxs$4XX0BkjDk@SBv$6L@ac}6@*#@wdob#wJrNqTL<%I<)PnO81&3Ad#VthsmK%Fgp;$Lezf7`#N zl$6(Zv@mT1)DkWz(RXcJ$pF02AgJWxxCFKE&r*8!Lq$!kBLT{2^=)W=ye63M-t=>x z>mFXyJmfP*tqI1oymA4ZDM|{ba&^V_=o?i#DTR)=XC6IB1tqWNj%vm%#HbD{WZWz` zOHVI4DXenKwqEej_V=#3F1-&=U+QRTX6Bym0u!Hd$~pl1ZMvXH2CvCzTI9rG{t!sG zyT&4dzf#lEwiXAR7kb|h`d;OmYCB+;q8y0kluz0<$7NgS4+S$S(i&9c=ojzocx_jP z9w6RsKtVF~JRkMaPSLZ%#^_-nHFYg5!OPh;^HvL=zp&WbdI#*yrz`R{H#fiTvlZH1 zo0~{UNdY!L?t>hW5*rnzud7Qs9BXQ6L50Q3IO z=Tr3e$Bm5i^c!=U_h9kK$w=4frb@qkX49PVb&cMcrstOhSpi(1XKw7--Xue62W9g< zz6D_WM;9(H<+$uTR=(`xMP(Je%6L8VBqm;lL(YjkvFNZsha7uzsxdP1@ne@tctPNe zvxPu%{qlsZCwH}C`_-~oA19GxVIU_G+vp^Tyc?VKL>U~<@DwOhL5-o3-{l!B3hXXW zx9-%4m|Dud|2|fQ#8KX;&>y`GPt_hDkz&UXF8AYdsjc`;yp-m&(zTNm#MlV;RRQn= z(H~+H+v7UqKf8t|p4}exlbehvg$;qaM?#TcqsNuqD>}}3dhlZc^5IVC(=exZ!@)B%mu~LL| z*#(Y&mZ~rJ!v~}Lyh7SV{`{q_y0I#|F4el m|M-{SzgAfLehqozG*tr1F7Zu%+S>6iLUk2wbhh%{p#KAYLAqN2 diff --git a/content/ko/case-studies/buffer/buffer_logo.png b/content/ko/case-studies/buffer/buffer_logo.png deleted file mode 100644 index 1b4b3b7e525d0a796046cdacfaec27aab25cc3e0..0000000000000000000000000000000000000000 GIT binary patch literal 0 HcmV?d00001 literal 4791 zcmdT|`9Bl>AE(4}6duInjtob58vEX670i!_>fb@=t-`mJ%IUhOBzFq_A-#g7z% z+$i=lSo1vEq`%O$$BQKrxRPrN>1X_qQGvMI06A$D$=4|)oG)aahkWt>Z{nxcCWz*VwwVB|G^`lw( zwC`GJaqMNM@L{Nt9T)o(>%pOl`whOMmZmW=K>IHSrkL#70k9&wD0tL>f<4Z}9*4yW zx&l~4{ox2~zjgzc$ znUEDCi{1i3v+o-|+`DlJ$c9yi%R|o*K(_YoMMHHSw!h#lVTYV_-#e}1xg6<;*&emF zsN)5Sv2s_C=Hr2o$I6fuy4oavD5X#mDVc5odf-0{(|MDNDTsT-lZkR~)FF3fZ#MrL zK+>gE+$sqncpxR0CcQsxlIaE+z9LOLIifzlKiW0{tVtHPn}S% zpCFU5Ox<;O$WeB63&?tW#>i!UKyOP+HK2`=1rZy+U1@XP5ejvCvm@vUB19bixZ1o; zX7i%nGG$KQ|0r}8-p)13c8QmtdcZOgf!9#H z(JX+}9ltx6c&HM+wNr41Wk?+Fn)~D|M6=1E%KpQkgTGIzBJMr?-eTM>A}|Pr+<}{? z2EKDQgs<~P7fMo|Jl)~Si3(AJ@GrNI?6hY=isgl{`&+ZK<+@T$XTR6)!trqn7<5+P zlzWH0N5#X^b(IHA6Q8fsn$|0QH|xfe4$7yqrThKA-L;*1Irz=OP;}th%Im;_$$;?( z9goT?T=%2$<4XN{Qa7Wl?`r>(75RJhd#ACW8N2?$1ZvF0AzLGYe$1O>56JIH*X)YT zGZ6}+!8BRy;9AVJTa{)ZPNAzZVJd4W;J3}Ey>+@l9UNCbwkEt$dM@(#CVip!RI~gz zBJy~t<0*?6Chs_DrFDPoeurCLgey{Y^PcM3TBFs^U@dYrvO_5+DbomjZzwCe?rr|h zy-cUb#(;a^u?aw#h(jhP7&G*ydb9q!{rUNrRp*-8YISX2x~z#PaYmEjhsSGWRekZG zQBn(i_pz^vMl3zKz1>jNxx2@wGhtgqw=cCD4(EHt2A4tsPt)|CXzDDVH2WzI-i(&c z?Zr~Uiq&X8yc&LxxlKj5bHjJdJ{j!8^o{uB2J!7(5|%-*fEy1qm=_~gv6}CgXbl~l zdbrv_gYP}<5o@BKb?!KDLsIYROiE8jo@`R5v4eGHw{cdbs9`gVM19@3<9Au`*E!|7 zQ7bQwCC2%7hHA*d;!BJAFv)t1zV$W};Y4>i&Dy+X10z=R?BUl>t{DZ`g|pE%z6Yv4 zh9@JT<;!+blWt_dCY=RjaR_nPAt4|(*xc|x|Ll{#bvBV+KY71d<4gKz$QhlHm8ITm zWmTV>LPV9uQ*W!d&qxSAvtI2S^7kzlE+|~k5VuagRfuxtS0&AzW7EXI zbK^LA2U_iPnL*J<#NINwkldc^kDOxR+wH#Oi*ksgtoZZ;(8mQt?u$YE6$90PST$Pf zpGY>#GI^FmXNMRGbp=>#v7=TZNlL4Y&a!ScVhNNft!Cb6i03_lmjlwAC7pEKeG@8} z8A|a7)Z4idyROS3E?EAFlypG#-P8 zSs&tSM$Hhy*f{(xbD&S7?_;B*x|5{)5XYLTQ9+9$6imaxSuX@}uEd2oIyOeHgZzYvuTY!Qm z-TDQ)bbUK25%po=Hdq^LTfSB-;_uz9`CK$`0`rh+aEHWO-?b1tb8R`RV~WMN%>ddn zBA4;H2tvJkfsBH3wQwZ_5+6}t-FaF#Xu`50I6u{`gnJwe1`2lm;?nUl2;QEnho`K2 zHm$zX`P^%wR3lP8f$TW4bWKvsM7-Ay;J;j@lfnqHh8mSJC`9`Ka}uSQ5_g^6STg^& zzAC_GxuYdAzC61>?qs+@k7gO%!o0eM77qS5Klrj|{Q?6dG{BZqW>K8F{aW7;y24$H zk}B`?gAe#dW&!fr=<9N2ifTMdb0Jx$a>uEG%ad+-`qOkmDYE$Ue2945r4<*2%+FC* zS(?+O#rV3kXmV1VcTVBbDpLeAzcKCIwL73v55&zPOEbh(1tvx460gCK(b#<-{oeOT z(XyxODW&9*3i|OKgtCsq-uO$Lz0-UtrON8j%_rn@cD!;)k48e}S2N`AZGImIvkVGS ziE2#ft?Oob4mh_#5JyM+-1J8gZg%v{F;un`@6OVEiMf;q$_OxMQR!Z&|L$G8&SxZC zJDtc?J4}vlif0?drFVR){=!>>eEJiBXIG35kU9yX3(8LKd&Vlukr2=w+O_Yk#VQ)L z{+Rvczc`-euzO`ox(<0E`2s@e=Pz}I>X%0X$7rzkjW(JuBO(ccyoaSb`TBrAWr=qH zb=E$8b@<@jkyGwY7$2YUo{6=sA7jSre|ZXj)OgdAu_U`II2g$C7@uf+bK4}iNEe%o z0Vos2zq@98c|ua-JI{{<2g8JLX3*t9uGv1#L5C*n=A2`Q&GkzfZ$$DSqH3y|mCKr>S$Z2pCKQv6JEHnq~u%Hu^2LT z1b024#JS2gB+-CDN6y1<3hZlIVr>ZF&+q-po;8ghqsStKbP-FML7qNpLnx zZj|N<4sfx3gwbC!s;)oj$r0gl(vW+%jGC;S4@Vkr`@LGDK=E;ye+9%EuNd7RiEeu- z8Z_;f@x%B=uohT4(Kr~to$(;hE?XRVTHm8kMgZnr<;9gxLeDKBr-5-f^0&+4lXkn5 zoLIfS#NIsXH4&y%SMjgCL(W6AU@uE?R=8*-657+`2>)jLT%n6#*^BKf_G{4d;s5QA z$4u6Q?zXdte+<9_TwG{HJRxx@HWC@i+Crm@mdswJCRqW`{`Fj~IqkPMzoK!oANr#U zR_rFog|(1YItq{k#w`B>v}p}Hq)}gaRMs$Ez=8>7m_pBz3;rev_TamFGs=<51ruRpl;h1xXzXbEt8z5xR zJ|F6Q4(_D7hk^UDQOL~!_-8(_NYzlWp~9^PXnl3D)p*59wT>4oyql*qiF_-~P*rx> zLPz2vAM*j7Hk3EzFO`O^)%;UhMmoy50MX9aA5#D6>@Nlu8RJA&)3~-ensj-D33VUT zk|7{<)4*k-+pEiq4q(jdOQ5gqtYOQ&S=E&fS)k2@X?9+1pD@kyZF>y79m$4gq)k)^ zUpGim2%5}jcDRsUn=g4Nq5LNn~ahaL!?7>>Pju<1xXdsTbbh1n*s6 zP@O>+X}Vl=S36rqfR)&uy7NVTM|yV1@JNMtur-PcA135moaSOlKDk-k!1bB!o2&Z> z));lOaiB>FOQl)yVr2O8%x}0UJR4nm=+7YJH~!UHOP_KRia^@=7x2pj;R&v6f6L^!t`4Y8eZVp_K)ZSaaolu;;7JDM!N=Me@iZ-9sk(EV>#PvQ+K4>qa{n29_ABHQ`M?4XsVvg zd*97woH?-Efb!zn!>1m5x%~jIXlkeNaIX6?=6;V@!_OR$NP^!|1N_K~VDnbKlRy8% z4!B$q)eIv+E-d6!51b~J2)x}sP^YNT;#!He9waxgE1oMpWfMefYZji56BV1U{gCXH zs6AG_W7(KKg{vVLSC0aK`LJZZDoVtJjbq5eu$EZyR(T$qnX}*)C5Kv`*V$Lk`oCw! z?B}5IA-5SUgnNyMBN{GRA>N-8ObFsq57YJBkNA>IgY9+?wBuuw8s9(NQ!bk^qw*xF zpOI6SVjSyYyImtWj6kQPOIsbRPY#B^NNaBAq8q+QyTqt7B&al%^ncrlzxC)vZ+lkk zHooQxNaNk)p+-r>JAq7$5md@6?X%WvCnzxEjg~jl9N-NJs`Y( z4saZtkUzESXRx3!gI$;o|M<`3uj-PxKa95?+1;z$9sT - Making Deployments Easy for a Small, Distributed Team -case_study_details: - - Company: Buffer - - Location: Around the World - - Industry: Social Media Technology ---- - -

Challenge

- -

With a small but fully distributed team of 80 working across almost a dozen time zones, Buffer—which offers social media management to agencies and marketers—was looking to solve its "classic monolithic code base problem," says Architect Dan Farrelly. "We wanted to have the kind of liquid infrastructure where a developer could create an app and deploy it and scale it horizontally as necessary."

- -

Solution

- -

Embracing containerization, Buffer moved its infrastructure from Amazon Web Services' Elastic Beanstalk to Docker on AWS, orchestrated with Kubernetes.

- -

Impact

- -

The new system "leveled up our ability with deployment and rolling out new changes," says Farrelly. "Building something on your computer and knowing that it's going to work has shortened things up a lot. Our feedback cycles are a lot faster now too."

- -{{< case-studies/quote author="DAN FARRELLY, BUFFER ARCHITECT" >}} -"It's amazing that we can use the Kubernetes solution off the shelf with our team. And it just keeps getting better. Before we even know that we need something, it's there in the next release or it's coming in the next few months." -{{< /case-studies/quote >}} - -{{< case-studies/lead >}} -Dan Farrelly uses a carpentry analogy to explain the problem his company, Buffer, began having as its team of developers grew over the past few years. -{{< /case-studies/lead >}} - -

"If you're building a table by yourself, it's fine," the company's architect says. "If you bring in a second person to work on the table, maybe that person can start sanding the legs while you're sanding the top. But when you bring a third or fourth person in, someone should probably work on a different table." Needing to work on more and more different tables led Buffer on a path toward microservices and containerization made possible by Kubernetes.

- -

Since around 2012, Buffer had already been using Elastic Beanstalk, the orchestration service for deploying infrastructure offered by Amazon Web Services. "We were deploying a single monolithic PHP application, and it was the same application across five or six environments," says Farrelly. "We were very much a product-driven company. It was all about shipping new features quickly and getting things out the door, and if something was not broken, we didn't spend too much time on it. If things were getting a little bit slow, we'd maybe use a faster server or just scale up one instance, and it would be good enough. We'd move on."

- -

But things came to a head in 2016. With the growing number of committers on staff, Farrelly and Buffer's then-CTO, Sunil Sadasivan, decided it was time to re-architect and rethink their infrastructure. "It was a classic monolithic code base problem," says Farrelly.

Some of the company's team was already successfully using Docker in their development environment, but the only application running on Docker in production was a marketing website that didn't see real user traffic. They wanted to go further with Docker, and the next step was looking at options for orchestration.

- -{{< case-studies/quote image="/images/case-studies/buffer/banner1.jpg" >}} -And all the things Kubernetes did well suited Buffer's needs. "We wanted to have the kind of liquid infrastructure where a developer could create an app and deploy it and scale it horizontally as necessary," says Farrelly. "We quickly used some scripts to set up a couple of test clusters, we built some small proof-of-concept applications in containers, and we deployed things within an hour. We had very little experience in running containers in production. It was amazing how quickly we could get a handle on it [Kubernetes]." -{{< /case-studies/quote >}} - -

First they considered Mesosphere, DC/OS and Amazon Elastic Container Service (which their data systems team was already using for some data pipeline jobs). While they were impressed by these offerings, they ultimately went with Kubernetes. "We run on AWS still, so spinning up, creating services and creating load balancers on demand for us without having to configure them manually was a great way for our team to get into this," says Farrelly. "We didn't need to figure out how to configure this or that, especially coming from a former Elastic Beanstalk environment that gave us an automatically-configured load balancer. I really liked Kubernetes' controls of the command line. It just took care of ports. It was a lot more flexible. Kubernetes was designed for doing what it does, so it does it very well."

- -

And all the things Kubernetes did well suited Buffer's needs. "We wanted to have the kind of liquid infrastructure where a developer could create an app and deploy it and scale it horizontally as necessary," says Farrelly. "We quickly used some scripts to set up a couple of test clusters, we built some small proof-of-concept applications in containers, and we deployed things within an hour. We had very little experience in running containers in production. It was amazing how quickly we could get a handle on it [Kubernetes]."

- -

Above all, it provided a powerful solution for one of the company's most distinguishing characteristics: their remote team that's spread across a dozen different time zones. "The people with deep knowledge of our infrastructure live in time zones different from our peak traffic time zones, and most of our product engineers live in other places," says Farrelly. "So we really wanted something where anybody could get a grasp of the system early on and utilize it, and not have to worry that the deploy engineer is asleep. Otherwise people would sit around for 12 to 24 hours for something. It's been really cool to see people moving much faster."

- -

With a relatively small engineering team—just 25 people, and only a handful working on infrastructure, with the majority front-end developers—Buffer needed "something robust for them to deploy whatever they wanted," says Farrelly. Before, "it was only a couple of people who knew how to set up everything in the old way. With this system, it was easy to review documentation and get something out extremely quickly. It lowers the bar for us to get everything in production. We don't have the big team to build all these tools or manage the infrastructure like other larger companies might."

- -{{< case-studies/quote image="/images/case-studies/buffer/banner4.jpg" >}} -"In our old way of working, the feedback loop was a lot longer, and it was delicate because if you deployed something, the risk was high to potentially break something else," Farrelly says. "With the kind of deploys that we built around Kubernetes, we were able to detect bugs and fix them, and get them deployed super fast. The second someone is fixing [a bug], it's out the door." -{{< /case-studies/quote >}} - -

To help with this, Buffer developers wrote a deploy bot that wraps the Kubernetes deploy process and can be used by every team. "Before, our data analysts would update, say, a Python analysis script and have to wait for the lead on that team to click the button and deploy it," Farrelly explains. "Now our data analysts can make a change, enter a Slack command, '/deploy,' and it goes out instantly. They don't need to wait on these slow turnaround times. They don't even know where it's running; it doesn't matter."

- -

One of the first applications the team built from scratch using Kubernetes was a new image resizing service. As a social media management tool that allows marketing teams to collaborate on posts and send updates across multiple social media profiles and networks, Buffer has to be able to resize photographs as needed to meet the varying limitations of size and format posed by different social networks. "We always had these hacked together solutions," says Farrelly.

- -

To create this new service, one of the senior product engineers was assigned to learn Docker and Kubernetes, then build the service, test it, deploy it and monitor it—which he was able to do relatively quickly. "In our old way of working, the feedback loop was a lot longer, and it was delicate because if you deployed something, the risk was high to potentially break something else," Farrelly says. "With the kind of deploys that we built around Kubernetes, we were able to detect bugs and fix them, and get them deployed super fast. The second someone is fixing [a bug], it's out the door."

- -

Plus, unlike with their old system, they could scale things horizontally with one command. "As we rolled it out," Farrelly says, "we could anticipate and just click a button. This allowed us to deal with the demand that our users were placing on the system and easily scale it to handle it."

- -

Another thing they weren't able to do before was a canary deploy. This new capability "made us so much more confident in deploying big changes," says Farrelly. "Before, it took a lot of testing, which is still good, but it was also a lot of 'fingers crossed.' And this is something that gets run 800,000 times a day, the core of our business. If it doesn't work, our business doesn't work. In a Kubernetes world, I can do a canary deploy to test it for 1 percent and I can shut it down very quickly if it isn't working. This has leveled up our ability to deploy and roll out new changes quickly while reducing risk."

- -{{< case-studies/quote >}} -"If you want to run containers in production, with nearly the power that Google uses internally, this [Kubernetes] is a great way to do that," Farrelly says. "We're a relatively small team that's actually running Kubernetes, and we've never run anything like it before. So it's more approachable than you might think. That's the one big thing that I tell people who are experimenting with it. Pick a couple of things, roll it out, kick the tires on this for a couple of months and see how much it can handle. You start learning a lot this way." -{{< /case-studies/quote >}} - -

By October 2016, 54 percent of Buffer's traffic was going through their Kubernetes cluster. "There's a lot of our legacy functionality that still runs alright, and those parts might move to Kubernetes or stay in our old setup forever," says Farrelly. But the company made the commitment at that time that going forward, "all new development, all new features, will be running on Kubernetes."

- -

The plan for 2017 is to move all the legacy applications to a new Kubernetes cluster, and run everything they've pulled out of their old infrastructure, plus the new services they're developing in Kubernetes, on another cluster. "I want to bring all the benefits that we've seen on our early services to everyone on the team," says Farrelly.

- -{{< case-studies/lead >}} -For Buffer's engineers, it's an exciting process. "Every time we're deploying a new service, we need to figure out: OK, what's the architecture? How do these services communicate? What's the best way to build this service?" Farrelly says. "And then we use the different features that Kubernetes has to glue all the pieces together. It's enabling us to experiment as we're learning how to design a service-oriented architecture. Before, we just wouldn't have been able to do it. This is actually giving us a blank white board so we can do whatever we want on it." -{{< /case-studies/lead >}} - -

Part of that blank slate is the flexibility that Kubernetes offers should the time come when Buffer may want or need to change its cloud. "It's cloud agnostic so maybe one day we could switch to Google or somewhere else," Farrelly says. "We're very deep in Amazon but it's nice to know we could move away if we need to."

- -

At this point, the team at Buffer can't imagine running their infrastructure any other way—and they're happy to spread the word. "If you want to run containers in production, with nearly the power that Google uses internally, this [Kubernetes] is a great way to do that," Farrelly says. "We're a relatively small team that's actually running Kubernetes, and we've never run anything like it before. So it's more approachable than you might think. That's the one big thing that I tell people who are experimenting with it. Pick a couple of things, roll it out, kick the tires on this for a couple of months and see how much it can handle. You start learning a lot this way."

diff --git a/content/ko/case-studies/capital-one/capitalone_featured_logo.png b/content/ko/case-studies/capital-one/capitalone_featured_logo.png deleted file mode 100644 index f57c7697e36fd06fbd418787ea15eea218bc7692..0000000000000000000000000000000000000000 GIT binary patch literal 0 HcmV?d00001 literal 11306 zcmcI~WmH`;*Cy_6rMQ<;>;e~emm(K;cXuuh7bxx&T3m~}yL<7XE$;4aLtmTu@y_>a z=B#y2lAZl*%g)M4R>GAOq|i}_P@tfo&}F2>mEX#Rx7QyD=IslklXmo0ymyt*bX9RM zcl9uGHiHr|buc!gkg+qeFjF=&GWB#CHsgnaf_1i1)pXUA|IBORV8?9qmxmc_=lF(( zg5nnfI~tkTn7L9In^{=d3s9Z5c2ZGTnF>&80OeWb9mUKnt)#u2%~ZS;R872WOn6MG zgaj%0!Mtw*?95z^D8P2M_Ab0&0jht<<$bIFC9_ab{DZ~SMu6&{Mrq0`QHVJ>n^6Fn zIhjlV9IO;vJj?)2Rt_M5k%EmCz{SD}WMSoGVg>NBa`UnQDE{?Oy=ikcHRn|pm;9Hm zx0V2vrK_tWFAIx@hX=C5YTQ#nax^2+U;fLj7+E;$|); z&Q^}DRu1+Qemh)*xuz|fqDxW3)sk!1;EV8VrTc)uYYj6 zxGJ0dZ!`WYw~MN$qZx~`nTvy)v&q|dm{b27{1&_aSMy%XyZvkXwGh;J01{j<1{9Dd{ z#RC9hJR(3K7biPA03gB5!zIEc&Lzq!$}Yym#SWDGH&({p#ns5(#O&X?t^Vr%Czj)X z#PW(cn;E$}IIB81*#0{Ulq?-w9b7CO94W-a{*F438Aw4dZ)9R+|Cjvn?@;~Ib8$0g zD|a(fNoNN;ihsnK*XqC6!Op`i4rCJ*XJ?ZH{`G-df{Rm3lwAbKF2>C+A@b(KKe49& zC4T?HO8j53EN{lJ{7sbqBXRysddtSY)&ER`x5j@al9~P6>~VgZCgxCp4^U9_3Nqp% zs^Fy~Jw!u1wFiF{cMI=?iJ9;?TMOFm0<1wVb1y zmSIf^XH5kHBuB&HY$aTS`#9HK%a}7Wm8^N>w1Ld_k#N`foRhfNh#u!95A>O1dS5&O zWi02TuxdVOuo1_(F)RJu)6k}fJ=f_u663B1x6jwFFAoR}#UCweQ53kGo8eN@&K{7^ zgMujx6Y-ECG;oz4EM1JFu&EHrq?NpwYWgVjDb^iQvF`(fXY%_+`v;2?647~@e<(D6 zcIB|8G#;x{_joy%p~p9w5c*14`3{O&Np*7F^!O&bH-@3~{=j;C|4OI)C6!1+4PL{c zz{hW70tZHlA|6c(6o8e@ffCP=D;-aTEi#e?FPz=yKneGznevHgBj`-34VszR&=3WN z@=QDqkrN*x0$PLWhY^JlsRjJ_w&H7i>0R=8HGO?p7#T^;K?EO`BG_16rm2R%1tSuy z9T6e0y)CK+57HZ(`&ENJ*7m!%)(^4^WsV0GB}F4L9hIsyr5UIUCEP_qYq$lC0r&=G znJ^x$L@+>@@Y;h#ozIke6&v*@jy<5W&miuu@haNKRod5NW2=}#5pE%1g8a;d1cgG@ zPFij@=6%LhD4&}iFRmjo3Koks^d?$Pz-T}4cUTK=M^3kj#_1oOy*~!@>T|3RdaxGd z-auanJ*Zq1E$AM#N#*F&4kqKcqrJ+9JUZ)&l|D=PVi-gv-Og&UJj9vMGu7;g>ah*gNRq?pOWFNlSFr zsM|ZhX?&jX{<@)B$m1En!LxbK9pnS0N-~*}I;lj|l^Ic}rWyF9Sq{sS#;}2=SAaPW z<5XByD;o}-wSSEQz;6}>iKRRieql(bS(8)JFj#WJ!_!bnQb3Ns%ZfJ%?R zThlN22|I1?;hk~NX5Vb!Upz4k3IKP6TzUn`kKu8n)B|r|ANBZd!X7b8G?Pm!QFEcL zsP}Y5oF8a9V-tSlBjwC~qn6Iba-1YS>b285pdR2>EHAFk;_tu{V*(`l)rg zW|5Fe%afWC8o;l=)eEZa! zm$ZB)Av;n;q*WJA0;>)sis`3&%`iUa=^1`kJd&}$78(LlyM8|X$}VEc`*gxNYjn*t zB|Ro1JL7Uriz(>ab)24Ub;@|jV09audR~3`@T7>Y7TsKKng&W;h-FYeE1XzZVzh#? z*TCcet4eNM2nB${eN<-4(Nm{PV07eL9S3}vg;cO`fROq)4@tv@@#92V7(_4dfJkw& zS4Qtg4q-wLvP*h{H>Ox4zppz6?z=)xBDlQFFhFT;F4EVCX-nw`nBXq)%w1ik?j^6B zmH{p0r7GhPB~@`wK4}`I0ea;^Ts769CVmhwbaJtzQKvZ(i4rFS#K5zRO-n|nlu2_x z;@n%_xW%kh0(H(cAP*snJspbgs>o;%2yshmY#oj0KKnvmE?Fw4koS1?0zJVMpV0@0;*c7GB0vp`)q7haHQ9H0RXR@cC6o)Kxjj88>2*sk1`I$HR%) z`tsF_jg9yGG?bw(%(2jUzQ^agBb)KEo}=rDh)bACU&pA)vT;Q%dKf$QS5slQ5Y(g!#)gu zPtU2MsXfPHI)TIana^VJgVI$szGZ8b>v-bXE&T2SQBn7o_{7y}K94tia&o|S@At~e zmS046*&h6h8Ej%cwcDk8JoD1E!w!JB-{<pi>rg|mm4#hX*OtWG$-8! zXsfXTbdVu&qZZ0R9G-?2b`;z7N_!Bu1;9T9n}vE^FEd~@l3rH13EZK8CS|`hLFEJ~ zjp9-^%`O=!H1i_kmEL#Vb&R85cDYCL-io7lhNJ6OiuzGrK>uZ5=^@HV1|nHewDWo3 zQw+IUe_y&zgL=#E1@@ta0U+rIPzJmH+0mFIjYbAy`f2nIW7szaZ)OU#`>>M4y|GOt z5}L;fOa$~(FwNMfigC-Rqc~sYRyOFmU00C_`AD(tnw%%~gr!9OYQN{iJ~PIP$eC3a z4@T>ca|-*)u9TB?J(rwzl9Bqld2ksd`sU3$WX*DKW%jycIMLBMnJBwM3OQ6HY2~`# z?G-z}?k`o$e{yGe6a_mlqw+XE!r{xrdyyvKIx6TZhirbqlR+pcZKs;|?EO+yTERQv z0agguGlK0ljN63|9YoCX1OKsHv_8mj)rQKB|_Vfbvru z+a|C~`a*f?R;qQb?5~gRsmo=9ySv;NM0}J3U98qYKdwK!JGhIyQs(`ek zR+G_XVBld^9%Pf!=559p=y1{}yGp(W+yS4 zdM+tbig}vpb)_H`x(vyyntQ$J{d8xdaY#%jsJp&_yzcc)$I;0jRa{;=&PbSa z1j&VineVry#~=m7o%)@)^^Rf6l(0d+82HD-9DB?xPn3nbp`5|-qdIl=`tAwa#=`45 zqm5&!?MMj7Y}RK2k1$0Y_mwUu-}%lT63N)Hw(OWsJ5UajdwTgWG?=sazJf9`t_Dw0 zWa&Pq#o{p0>(amh8_}A6><(z0IC%Jrh}*cQtlaDg#mzik%y=laRCo60Cqf@yRT;eM zazBTAhycD{uxoU3mOCh*tWH*yiZ!U07XCEH_M+s-fRrY|SqqaQ@;;xHoArgEE}(X5 z0N;^U`k$T3dD!!W{UXA%qxh`aW=mp+-}YhfSVn9Nt2xge2KlmO?!O(PDZtLNM(IHJNa21@0w6PZ9C5-^) zS#AoMAM69L8g@Ey^4w!JUsN6c?U&o`HQw#uN#BI!(^{4&?-WFXqust`>?C z6fLMTh`#3)j%zlZE8))+yFC19UKy}9^krSVrz3kRxRe`E*s5msm9Nu zO}c{qI3lRi9?ewJ=CZR8OWyuW6=8oSDxbK#y&0g8FgEq!eI@m-&(+LBySS}t5fM6f z=Ik7#GV?ei!za07T!)T-u;DIF|5N2x+#r#y8+E@&7%-XL?H}*Dn-RXOP*|Lc=@=%z zh!=xVCw=`Er}0CB?kW+%Ebkc|2GD|4%zSZudTl(g-}TYzYqiE{?Lw3wb5&y*??4fv zG9GwkX@e{|%R9C0j@ZhoM=q;UzZ%s|8g2JdR?euq%ei)FzhgJ1)IzbXL+ogH#N$u4 z{Uuq`cKd0g+X%POdt)UmAKS_O^WS=8j)X=ui_7{fqX}&+uUiX*&EklNH*FL> zo(D(v0b!$?tGcS;XcsX0!*fs@YzL7H#LAnwB>J5ZH0-#>qLMT$;KhL6;~+$*N6nNJ zH?QZlu^vyuB+31{;pC#aCPMBw>rDtp>r>eb>=sGMZGeMZw#jF@XJp&?HCF?ka~|7K zwr)t{t8~^tZq=Y9X6EAx3P^&Q_;zRHICj<9>>+ADRNo(XpR+Gu;1|#TTLw){UB_OQ zaWHRC@b2YV%AQM#n23a_IC;;dKkT$}SzzgNy($IWhdOxPbF`cJ4&?WHp5O z+cMKA?uD(2r|IRFKP1*rf*7_jCl;-b!c4NjrbjLw8TR|%5m#PL5+~-N=aqQ}E|*b@ zs&hPbM2_qlGf?|COki7MRJVWhq2^})*x!hF_{3*Xv_DgmKd~U``q=z_!WWY8Q#q$y zNto<1FiVWgpVT?ktKWw#T`je{ysl$XhewnrK%6v{E= zbAUE#xd#SGk*S`aZjwxaC=miV5*-nL2r@qv z8RnabvvYXMH9s?CY97b)h|ub4WT_bjU{lZ$-e@HS4G^s);^wN_f4@Ej^G%!Un!Fix{}OdF(xclvv>wo%dymm3vQn z%5qS4GC!rBmSpc7QOS-yqS5EBFF0qJ{UhYbNNWG*d?!vlTK8mAG<>j)J5PHhBEzM>%VKyG|9B2glEow5 z8_qTWRP2d>Wg7EWO{TRn7p4WoJ=@rh?)wJ@vJUbLpzfD$d09$tG`~Oh_qnEkPsNS6M3&Qsq2GU%ufWA?=QO} z@cAM!o|+EF;A89HjpY+-H7r@OlE zrL0rGtXytzB%BXD6fAx-d-1P?y z`&7+kR;}NxW%dw#I~@zP)8kywP=4ecFq@u_ZbMx!aZ_lhZvSn*h*;f0h00BNH z6?t2=fpXOPjn-*cGQDWS$$A|9^9K4ORG3d3qJkJftu7JPE5=#1%n!?UP8^cEiH=^R z(A%N7j1NZiLYKj2wwL5B(GU3Zfe8Z>Rb-3wSi)^6IxWsT71@(qh}2!=YjP!JU17nt zAB|F^;8GRQ2DR47cKjMJ@OUq(n)Ued`!lVcfVMfQd0iCDg8H8MCFg5_);`3@I{?^L zi5s08mvwUUb^(`zF?;k)hXn}jlVCQmB^M)20S zw|?{edPvMGuXLLV1+70ARe?Ls73q|nfgT)4+i@Jfis#oP!AFMBSbdisT*1WOEQ6*oE2~HKvL}nfU;k(CjKTcn`~9ZfQAs7JbN4Tz*_; z{+egT>vgS~TL0pyEN0k5s>A&hz3TjH0dbb^x{JwfUD%HA#xp*ukb%q7hbqk|s18hx zQz#)@!_)Kz`lqAap2rifk1dMs_X*8*4!I_f_x*%M(rO$F7t|E71-E1J&*<3ZN1cgaoMJ`u zAxJ23P$b4l>PO5RD+eZ>4y#U!=hf}FIC9C>u-nc8fBh}~(p!>_(?BjIFWvoj6X_gy z?kQ>}c~bJkb)5p9=RX5112D3s@roZw={NzhufvUL25CMfbuX^$HbkT+{yv@e*AAS> zlgclKmQ!$O29_a)P~I68Z%e|(B51-G3+ts9%TWtY}=YHRK9*% z7A`;(Y7D-A`!eOF&Dr~-v$rEFxr#QH(A(L%Sum({On#X)@Mzv1gSbxKLH7 zT*yRhkiY;Cby5M3Nyh!cmef>=f{vX4_l!&3z2qk$s$t{ybVvSl{-p{MdqPFmI@`d{@G^FL4pV$Q;+%u@|W9xr<^p4{+^zPR76 zoqS(&!?8=E9z<1>XA;b+^ftd(F{w{1WD(z&Sxc4W)xh(8;ZF2BevEZf5a{mF?0nTE zZVHkc6-#l2*#*s6vQEk{T+tElNd?)9iGZxsOmx_{A0fGl9e0=!kydJ>{Nqp$y|e9b zB1M(v?(+GMCXWmpVf;FaEF1<-U4Qo-q{xZAi#fon}!unP?r#jul?YOnu7fsesdYHu0 zRP`B)9Sj&frr&Mkh$`*2#@fcRj4{he>c5Osw+jWH#yvP_$-uW0A4spqOGt&MVJ~XU z6c)90oowf^33JH>+`V=3&mpYa*R%%AnxgK-GI(e*=b@*CK)-p@9Bl}_1JiCy}YjHa(x~`xtIL`dsc>7GHz1SF^At2$+B0994R;kjrm}wN zvSWag>|=LuUczIZgVq>U=xKvt9twmgC&dS3IDx3-Cj0%|Pui)Ma$aEqyu zqWUOSKYWtNBt7HgtJY8m4&~T$&0`+6jYt<6n>51J;-0x}4jMbIur$w%p~2(Ax0xpm)i-64gLR!*GpB!Ufy!O^aH`vomk!9Gx%wjfoxiMkqFGFOtX z>S2UOCI5F(Tw|Ve1<@e-ex^hnfVmq+805wC`J*hu*YI$*eJ94^x)Nx14tbL?icU-v z=KT5b`=dcElhA{DqGZ&gkELE`<);Uj^Vk7;096_l=Id2<#0ep_?&m6!xg*L;tWbEu z$l+`&U_EYE3%si`$AuTapTc0eNQ!@z`G=Uw^M$RZ+m;TtsAPP+UWw4vU;g)hh6~AR z9v$(W-~OTUXchZ+Ztzc7#0U~2Q5jdNes5o&&{tU1HQvE)cq=NB3KbphPh_M&=ZB`# z$n}=Yznnb{9{MjaS-XL6q7p3B`s2u9#~JfzwMRqcTGEWzHXPAxALO5R#&j4CR0^OO z)=*Tn5qx%u$+QV`hY=7rp6L0a*$#q*S&u3s43n1s@M)NLmlxnKT!sIX!o{HsO*P6(K|sMeHn5mUFye z>DnEDgi}8Mc0ij$daKTVyQp~BfKSO8b}t!iU>3B8^g^RY_dXE(=Mi?8pv);Vi(^CD zZ25N89v{eCc30$B3{Alria1mzR*7-?bYF32>AT9O-vz5pTUdZpw5Yb(b%GQp&vXUI z#HKtqMIsD;F`Q`oCDZH0f3w5(>!`~yzi22;E-ppP)bT=Zz71efS0-+FX>oX)O~8WY z3$VS&Ch%8Cv)tqQjX~9lIN^(%lZ{zYf8Ask@B{`R9PePh@`9)3ihAHvaIM&MuhV<< z{qxY%g8-O+E-rqhWj;DTZ+K|rNK+7<4DMUxQ?VI9pe;P8d17&G%%NwOE|nEnK@~^o zfge`yDLDEsNc(+j6+J|TiC69VhtHsNzj)a!l~bDOIRteF8M|6J9jn?K26fXq%E7dl zXeySwYH%u{vJxR*+=$#G10-F?w>-=WZze`H?>C<7-8Do&{SCP_kc5&=c_!wpESJYw zuX5RtUK|b?>-Y6BG`t0a#6Zo`p1I1*?`=29B0{!I{HilYXLp_=Z|C!*IP$Q39wg~m zGqg;ZROi2wD>{FaMyLWKk`p6=jarv`VJwx}8m@t1WDajTuuv&6`N)l%L(i^{sP22?JqXsX8>@4HL5>$ zE&eK=34qo^^%I2p!|8Cow73}xCdqY zc2{!C{~=SwYlODyr{>`)=hgOJ6VKMG_4lofaBy;Qb$$LzN77?$GZ@9Iw12Rg1abM) zEC?osSj8G2-W<+HDgKzie&AC8R)U;l=A0{jpzr#|d2Cy_i*vXmQdPr!f9L0KKzC^{ zmD1Jv+V`Tj3nZNnJL#yppg4kxUM${gk);lfSfqC;l*lm8BUFqH4L6hDe(QOA3@-C> zdJON*VkpkdvV&w{YL;I4U4Ei|VE$osYwk?fuIi1~X7b8jd+KA|Qy*Vh;p;V(y#4zz zdawPy-l~wSX|0vX-`5YY*LvWvJ0LToF+cudI6C0lH5mDBoRBN~xd1hB*Pka>v5-$8 z{3ty>eay{u5_*c)O|*_uwo-fhFj{kbFWHbtrri;E3y#Yu*-m%`Dd@J9WU71vUBfpYry3TcfMt3+~o7D~$ zvS-vxtAqJe_1wkcAhhy{?|lcK{)%Jok;K?qU z@^y=E2f>#6l4UD@ljJD-xX{x}(sw_%%4xsmVpwG_fZQ^4_&D|ag%6~m?>Q}aTCu=+ zdCdZ12itC-Z-%ymC*0nj1G;*^)zY40N3s4(x4i_WzXW(qj*eO@*fZEDUw#z|jio(9 zv+ZL#_`qWD@9VSxk*aDkMP>*TVKwzKng$>E^bZNGW@IPzuih+G+gIOW{Ew% zW>bH?{Y-qhQg&{!ea<34F~`BvD(|Z{#!AEV0f}mE`!FG`&2z;SLcuIUWoymLZ&xlGXm041#SL z*!!(0sM&fuInhZH6B+AwmmFBg?1^1ne>?jwavq{_)m8qsTOq{#UQ_DSMDBlhxZRlq zPn^LmL9NZ@V>o|5Gw}2!sq|Gy7%r=8=O*YKj)-2syDMw$ybid-yRclh;c1gZxVmSsDpb0Tk&#~956$j)!YKn2nHdj-LQ^16ik%JpY!@>eCW`1->nZu` z%;&pwZYSlXOPB9kTcd6Acodlq=Xm&}B&?Hz4+#|=f^O^lzPtPde>*hRwQlb6VpCYW zR&AqBE@rGvUO!D0?CX6LZVN8h9|Y1i#y~LJQyAG}+xFS~Ju6}QP$_JEpI36kd@lR zU4yz=H9dMR)VUAp{G<}U729du?TUW(@EC|l9QutFK*;~HBb2I9HspEUUj+U972)k( z52X?z5=kc4MBHNa=|jD&?#^e4?;CvH!hMnbeJ;0DS|Ql@JnaB^kSz6+IjDy#f?ilv m92-9D|9h#ce&G*;2E~F9fcpCe`TXzgT^R`l@d{Bx$o~Sc%O{xt diff --git a/content/ko/case-studies/capital-one/index.html b/content/ko/case-studies/capital-one/index.html deleted file mode 100644 index 7474b71b10..0000000000 --- a/content/ko/case-studies/capital-one/index.html +++ /dev/null @@ -1,61 +0,0 @@ ---- -title: Capital One Case Study -case_study_styles: true -cid: caseStudies - -new_case_study_styles: true -heading_background: /images/case-studies/capitalone/banner1.jpg -heading_title_logo: /images/capitalone-logo.png -subheading: > - Supporting Fast Decisioning Applications with Kubernetes -case_study_details: - - Company: Capital One - - Location: McLean, Virginia - - Industry: Retail banking ---- - -

Challenge

- -

The team set out to build a provisioning platform for Capital One applications deployed on AWS that use streaming, big-data decisioning, and machine learning. One of these applications handles millions of transactions a day; some deal with critical functions like fraud detection and credit decisioning. The key considerations: resilience and speed—as well as full rehydration of the cluster from base AMIs.

- -

Solution

- -

The decision to run Kubernetes "is very strategic for us," says John Swift, Senior Director Software Engineering. "We use Kubernetes as a substrate or an operating system, if you will. There's a degree of affinity in our product development."

- -

Impact

- -

"Kubernetes is a significant productivity multiplier," says Lead Software Engineer Keith Gasser, adding that to run the platform without Kubernetes would "easily see our costs triple, quadruple what they are now for the amount of pure AWS expense." Time to market has been improved as well: "Now, a team can come to us and we can have them up and running with a basic decisioning app in a fortnight, which before would have taken a whole quarter, if not longer." Deployments increased by several orders of magnitude. Plus, the rehydration/cluster-rebuild process, which took a significant part of a day to do manually, now takes a couple hours with Kubernetes automation and declarative configuration.

- -{{< case-studies/quote author="Jamil Jadallah, Scrum Master" >}} - -
-"With the scalability, the management, the coordination, Kubernetes really empowers us and gives us more time back than we had before." -{{< /case-studies/quote >}} - -

As a top 10 U.S. retail bank, Capital One has applications that handle millions of transactions a day. Big-data decisioning—for fraud detection, credit approvals and beyond—is core to the business. To support the teams that build applications with those functions for the bank, the cloud team led by Senior Director Software Engineering John Swift embraced Kubernetes for its provisioning platform. "Kubernetes and its entire ecosystem are very strategic for us," says Swift. "We use Kubernetes as a substrate or an operating system, if you will. There's a degree of affinity in our product development."

- -

Almost two years ago, the team embarked on this journey by first working with Docker. Then came Kubernetes. "We wanted to put streaming services into Kubernetes as one feature of the workloads for fast decisioning, and to be able to do batch alongside it," says Lead Software Engineer Keith Gasser. "Once the data is streamed and batched, there are so many tool sets in Flink that we use for decisioning. We want to provide the tools in the same ecosystem, in a consistent way, rather than have a large custom snowflake ecosystem where every tool needs its own custom deployment. Kubernetes gives us the ability to bring all of these together, so the richness of the open source and even the license community dealing with big data can be corralled."

- -{{< case-studies/quote image="/images/case-studies/capitalone/banner3.jpg" >}} -"We want to provide the tools in the same ecosystem, in a consistent way, rather than have a large custom snowflake ecosystem where every tool needs its own custom deployment. Kubernetes gives us the ability to bring all of these together, so the richness of the open source and even the license community dealing with big data can be corralled." -{{< /case-studies/quote >}} - -

In this first year, the impact has already been great. "Time to market is really huge for us," says Gasser. "Especially with fraud, you have to be very nimble in the way you respond to threats in the marketplace—being able to add and push new rules, detect new patterns of behavior, detect anomalies in account and transaction flows." With Kubernetes, "a team can come to us and we can have them up and running with a basic decisioning app in a fortnight, which before would have taken a whole quarter, if not longer. Kubernetes is a manifold productivity multiplier."

- -

Teams now have the tools to be autonomous in their deployments, and as a result, deployments have increased by two orders of magnitude. "And that was with just seven dedicated resources, without needing a whole group sitting there watching everything," says Scrum Master Jamil Jadallah. "That's a huge cost savings. With the scalability, the management, the coordination, Kubernetes really empowers us and gives us more time back than we had before."

- -{{< case-studies/quote image="/images/case-studies/capitalone/banner4.jpg" >}} -With Kubernetes, "a team can come to us and we can have them up and running with a basic decisioning app in a fortnight, which before would have taken a whole quarter, if not longer. Kubernetes is a manifold productivity multiplier." -{{< /case-studies/quote >}} - -

Kubernetes has also been a great time-saver for Capital One's required period "rehydration" of clusters from base AMIs. To minimize the attack vulnerability profile for applications in the cloud, "Our entire clusters get rebuilt from scratch periodically, with new fresh instances and virtual server images that are patched with the latest and greatest security patches," says Gasser. This process used to take the better part of a day, and personnel, to do manually. It's now a quick Kubernetes job.

- -

Savings extend to both capital and operating expenses. "It takes very little to get into Kubernetes because it's all open source," Gasser points out. "We went the DIY route for building our cluster, and we definitely like the flexibility of being able to embrace the latest from the community immediately without waiting for a downstream company to do it. There's capex related to those licenses that we don't have to pay for. Moreover, there's capex savings for us from some of the proprietary software that we get to sunset in our particular domain. So that goes onto our ledger in a positive way as well." (Some of those open source technologies include Prometheus, Fluentd, gRPC, Istio, CNI, and Envoy.)

- -{{< case-studies/quote >}} -"If we had to do all of this without Kubernetes, on underlying cloud services, I could easily see our costs triple, quadruple what they are now for the amount of pure AWS expense. That doesn't account for personnel to deploy and maintain all the additional infrastructure." -{{< /case-studies/quote >}} - -

And on the opex side, Gasser says, the savings are high. "We run dozens of services, we have scores of pods, many daemon sets, and since we're data-driven, we take advantage of EBS-backed volume claims for all of our stateful services. If we had to do all of this without Kubernetes, on underlying cloud services, I could easily see our costs triple, quadruple what they are now for the amount of pure AWS expense. That doesn't account for personnel to deploy and maintain all the additional infrastructure."

- -

The team is confident that the benefits will continue to multiply—without a steep learning curve for the engineers being exposed to the new technology. "As we onboard additional tenants in this ecosystem, I think the need for folks to understand Kubernetes may not necessarily go up. In fact, I think it goes down, and that's good," says Gasser. "Because that really demonstrates the scalability of the technology. You start to reap the benefits, and they can concentrate on all the features they need to build for great decisioning in the business— fraud decisions, credit decisions—and not have to worry about, 'Is my AWS server broken? Is my pod not running?'"

diff --git a/content/ko/case-studies/crowdfire/crowdfire_featured_logo.png b/content/ko/case-studies/crowdfire/crowdfire_featured_logo.png deleted file mode 100644 index ef84b16ea06c7de72fa288068f309fa91eaa36a5..0000000000000000000000000000000000000000 GIT binary patch literal 0 HcmV?d00001 literal 7827 zcmbtZWmH_vmWAMrYoPHaSa9z~12hhey9H?^NJHbEK+xds1PKI4LXhAR+zG)#aF^ij zVaWI1_hx>)H8X4Gu61wis=N0&r%s*ot0FaE3itpZ00jjFUr7pcWs1?yO#wLjv{T1u!J)xIa%1kwcr-kKCb<6aTF9ZUwa(`cLOz5Q7eQK zx5Xb0x3`nagBk@zT*}+U!pZ^e&R_|*wRe^P9W=Fp80@VjK>9*zylO78a65ZNUnE@H z7p7z7>tF@321!XWh82+NTJ4k^35tMsS5v1nn5@Nky^Jw*DCEjYQTsr}2av)ext^^h_iZwnV5K5kwf zC#OIA`it%6t_A2L8v?*4m6e~1s!h^iy)9}dOB z5sI+#aDqF#D?ued4}Z9=?X5)xt*pSpyn+^7yu!l#T!Ij5K`skkF#kgWdHLb|mX>^$ zaLC_w{+(U`0)`6l%g71v%M0=G$$`OgBEqsV0@6YPvS0x@Y5BjkO3rTX7S2}izkS<3 z`2MFB`rm3rWsz_TcLY)gfpGl03N-8x?g%$KgbRbL>|dwOEyTd4W?^OT{AZc{&r$tj z=TJD(-V<&uk3={z{FQ4_`+pP1|1Jmr&|CjsbI9`$6wjZ6`7Z_Z_tHaE{(=8$hlk0( z8Vc_G&_&3Hc99^Vib6pFRwzNGb-ZVGO>xq7CM!RQ&R3s(!0W{8l&mU@2!781Enq09 zM|&DMJ5Kqy;H|j6iJ68D-qNRMeF9H`Odu$+;7jbj*oeMXKo#K-O10NTslLZSPD+9y zWl~A0)Lo34*WzCG{jF|LjvM0h=g&lBQ9&Xak?%D?Psx;_p(L954=6bT@CQM4*d7pb zfy^I7O#TNw5FgM3@elexNd71BZ#5`hC@oSG5uq-926DL)cd&n+cw@1@Ymx%YY+!)|o=)Ln_|) zw1p7MvpUXTg*@CKek_utc-hZv3{mJtXvg&KY}+R}vp9`4JR8guq{MGB?B8E9T`|;Q zH+?A?7kC{DV75%*{i(2fKwsl9C5x0mp%-NYhU#)-v;+-gHOmAt0tqH$pr?@NPD`y+ zKiP1%yL$rhWN#3)kd+&b=(-i|>SU}cr= zk2$(ru*H11NkKC;@VjrhfOp!KmpM{Ukb}6~#zK>L-jio<@eMvOg4R`vOsPhOlSdt$ z4KSlY*@2#rKuO+hu7pj&;S?Hr`{S*wsy0c9DWdaq(o7Si7uq4?@6&ABMEC25?VTIj z`f+b1>nB5QY)g~uhSs)Y#P&ZuRSW#bOp zngxUQ07G-Q#0f+M(#nlaiw(^h7e_LdjIW8IIYHv2v z>C&JSHOMmXRe&>I4`x?(eKV>^td``UN;)%tW|t*{9z{VTK*` zoidE!2`*f(jMj)?BivaZRDBPa4Ip`dW5!*bSpUGo8%D|g)8^z)jVT3IYol#XRDD8P zO+h#Wn^QSK81RUQb`Xt7_C|oJFy}o1hQ*QN&`1%M*-HtrEky!h2ZDT%k+}Zp82yoe zX4f~#@Lf(-MU-%zHG3-5YF^2yX2B?!lRi$s6*=v;C}rj3C%AkFM*E6?g7~-KE%}R? z>wEv~D{eJkm^VFvAJ*xB40^_m-4BXvQQ~lMr*92JmX<+~D1&&`!dT;}%~kogNa9_q z>r>MR6?7J=YF)+EWDIas}x^hqp3RKm~U-340fbp>QB8>+m0s{s&Q0wa1~%6?3bFvJ>G ziJOLKLx`GiSl2eCKNE3s6cB<68Ce*{jN`Ei_k{znzK?G9UEpRM6Eh3&=UyQUb*&(; zb#blVH|fvh<5rpRpZF3cpF1z~G#?WC9}^S_7wp_03h;4o8ktvAwaY{wrI8i)XF!)- zZRBGw`6vd8W5$4wpE8vAot{EnUwc$TR?VUnO~O4Vn)E+%VkCi~`({x)V1_SQ zz%<0+w?eyiJVgf(r=&ko(=EW+@3m2m{CCs?tniCf{pzi~%9_BCgGS@smT$xg90!7& z^@V`Si2;`v-B*6%3N@?0eqs3ZJ>eb--!f1x;e#Wb-w9aQXy-UkU&ZxqW~A@dRPRh@ue{CzHVkn> zsF0*yn=wZzbk}_rKiL=?ND@{;Ht3S(t9U+qy!k{yT>17Zw{geg5NA!J+lp)y4BUXY zIfU9rWKbGv?B~YZnSNS_ec1G!S|ITlaIm$C;M(`BPBxF#b}Ow`7YQ`=OoCKnabFLe zw|(D{0aYVd{jYGKayTup0}9D6<^J_fwEkCo7i+o(@$8e z2DS6u)PIeJf(n#Wj^ zqmq_UycikQs)b%X8AW7}O3xgpCIQ#+7qzqd4AzBI`yvJ)Z|U5FovYJMPeb99G_*U! z=g715Q3s0E-MFoZYP+uoD&Y#?C#=m!)F#!@1ba!%QY{eoz3vhUvi!7_8p)TyokA3<2Beg;+1&$J)^#W6dJpexDoc+MEN4P2(i$%(LekPafG0mjMZf^}8>W|x$x6iuWGf!(w?>~~^gIBFf zmj>@Gf$QX%W@$^qolYeu;+-BY>fXC|GGB&3^d&`gl3K87mfP=0qn<2V3aa7}p}I6` z-XNfSImMeRF6=CUrZf`wVJ+WN`6dJ_=1M3iI^|?L$edTguQQ#na@}6g3s?`NCunR? zzFfFb4M)F<-|p5o(!oy3{X$b;Xyx*ic647gU0|~H9!N&c`>M#%by?`E&*L{cc?v%f zsWR?lG~9v`H-+EgR_WfVvuY=F_r2$zNq&=LUL=xNQXmr@<29~f^d+99%{fz~v$tpV z)8kw8sCo|p(W>xVJFy@cIh)a!NoCiJKCue*293x0nBY{k^2f6N5;n-*Jhzr-eG!9>Fql8D|r+n^HLdxnAeq6Tb;7wZlchZ+MxQbOuT z*qM2j?P7IKV4PVpoF=EpFT!aba=YgiT2WvoGKpW6qE|{pK*P z_EUGQs}|MQl`GUhQtCKy@q$9e_Q+iJ?LKy)8%Rc2j8`@Gx#;v7+1wX$aCxL8rkHT% z5ffL51pZj}`k}6vQ->sNxEWlZYfXZo4y{g!?a3wuc}Qe`vLD%gLv~KQM3dJUZrvmZ zL_4isVAdnV?)*K>36MLUGSok@R-e!@l8NIrZ_t1H+@r2XZINh)v@fFfx^&_w+b9>? zYi*a1%T}i+LlE&JrO_Nc=1JL#RLQ8e7eBoF!eze?IcbYhvx`zr(UJI*hALm0iG^t1 zOQ`jl{mX%>=J)9qo}s+<044Kpy=YDBolGFdMFy6BwdQm|T-&cN7u(rs#X*5}*|D_5 zGOq#4LnkJ|NG(Ms=VaWqHPxPHj7Ur0ZG5EUhhA#H-WY@*c`0Zxd5Ogbn_9lOxh8nK zD_diaKOK*Uq^KQ{comKwTI!)tQEZ@_HCoQJWmg-)mUn`o-!6wXuNYAw)xQ$%zpT1a z09uw!qeGHS+0(uACkuKKQ%|*Mti+DLBPe*Vz+7%y6tTyV+n4kqJ24R!`jWY|4?8y;mLn8@=H=v;=z30s;k~^ylV;7Vx4&L*GxAGLR(@v7AfF`;F@n)6fl_%Z znTZd`{Xw|zGlQ%+*(s0?o4=L~vEQ`$P6rz_^CmLOV%N=A!R&0_CZdvSXK*&*+L(Vg zQ()Mo6p&21R9}7`h&bObvIKRke0%lf^+k5_Kt^g^e=wEj4W_CSr;HuWfp_q=8OK!D z<>AVXYp*;|Qe)!1wWP8g*2K}A%bWB1pvSHJ!NNftzj9~N=>kL3ONaDD z{E(ZwH%DORGu;?A>iSsEy>;sOBjzp{Rly#4*$}7&1S@crGg=}M4RRHqMd*XUnk$v! ztx5B&)IuOC;{Dxw?9k7vkn>2XS$Qfg%#~>}0=EIS*4I8g%T;^Zr%P{APXg$UI6EtG zQkgAIZ~Er7XtT1=~4Qkuwo4VR_=+L*&o8Dw&im$0|Q;Tw#2IW4R zjy;7>3B*f>@Tu50k}tk$iTDBXLb`S&^;Moykt&$0+BR{0n5jx7*rh0AE_$AJvTi&w zWYg+klDR*_X*PL~rn2RdYTH@o^peZLTCqNbOfPL!y;(tOZ7$hVlliP0yhId}f*D-<|?zAJoHZHA8uY9#=d(<04~U`zhr|>cM5`Ppr@*=l(dfN9IV=u#F9_ z=LgPpLn}jrkWjWxB?rmwt_fiv5St8<6`-RRlu%7s@b;@Atmo=|`$Q4xm&aJ`rz~bS z$86@pScrgn%n-Ceoo{ad&_j|z^8RVOv0rjBkM|Q>uLiIRV8Oq-3HX93hB9Xb{BSrWFeJ?$9rTSwqva zS^BB&`Q6$a9z8i(UF!Cc*i2K|1SzQ37Wk9tbwzs$w^02qYdaO}bso8vmHX_LR4)X} z%s8Uj8(4359!^DDjJxVcf2ogIVSG3`U}uaYw*K^UO@Q~U%QLo{0a0wjjFBA%$ID{a zph&4bKj{j^BKt1)MQ;SQQ7yDsBWFWQWip6(M>$JO0y@j~=uvLEgG}6+bX7um2<)T0e&$IXPG+cC^&k zlqA8Z)&{ZTk!9>@GZ}T*j&-p5bq9t-&EMU>3wr)+gRna##0&9ew^n!jDJ*)@?vj$y zQ>aMT;^Ns{DSdRw&i*ICx7D1>Qy#TyC4e#m=Z%lIUd6`d*T8w)Wt-=3=} z)=H46Bc6QOX0YcmHj?c=NgSj!FV ztmy!S(BVJLuQsxNv$P0N%r|5=(^%$kr0ROMhRKN5yUy_mBKBasiQjBT_J`?x z7aMo#NfJxt>;<2Ue46G4JAjMs6bO?VHUK|_wy9=nE02%6Bi*`YBH2{D>m>Qn`o-d+ zgK^ZU^m+kKRn8bw{yuJBM#AO`AWco!Jl=!GEx6H#om(Xi2fFYoMr(R%e56-OGNM_u zV0U5V%CzsG{Ce&-;c#xz+P`{leZvCB=Pu}QyUCxvEz(V@1MW677w50q+QeN9(XQ0A z^4%cZ2oZ@F7%bZ{R-|ow7B0Yi&S#2@UFKjKit^c|;gkBY`_rN|9TO2l%y`O|5o?j> z0EVUHQkGA=oMjj3jhTSyLS(dLFN`XDYqnzGxx9!AE!8W^7-AjMU}v ztA@aCzD-6q!-^LK9J8M@e9p9i$WlSEa81366mt@VA(oQo?s&AK&#(efYO|eF@}zv+ z1Wi2Ap6g8hprz$iZ;=!6OuJSz5qB~sBs2+Ejh*QE9MG~pW$R<5>^hPg5ADq5VHKds z2?}7og$q7m@$r9Nn=@HHVgE%84S+kBB6S?jK7ErJ#X5RtIlcqP6fvdvCj=jX=3jB?y!@JBJ_7d1{f^L4mM zT6VvlB#p%fiba%dL&2W|1pOVVJaJ}3tOVe?=l?0yE|EzBnwr`J`^c->^qurh+ zL(UF2gbY~v)3`ethfG1faI`jc6p4_ZCH9ASZ#NVg)&$k{;n>E{S=_Qn#rY908cS=U zR*>i&SEY|t*7*aax@;!($B>ov|$v0{iDfsf}qgtI^KOM6DMH?8Q9+7ESj!DOG5 zj?!`x>^0(HG>RD&DI}}l@?v+SRKqpWxWEnU7`_m?Gd0D@R&mMVe~)dED>jJ-3mtPB zX%;^@Jou^b!?CDXq`D#~OkFkkN{=b&`50(Oioj_XZF-xPWi@Z1jK!8fIO%2&E~4`y$_e>2I&kr{(Il`l>qL%^)(7viW>dgxT5l|&7mR7zNXGR? zxfRFX=IhOmb?qBa-hEV-yg%CQCa6a<|4lg#6bp}a*`nOxgq-vbZ>KGyYfsTg@vC=* z|B~949{BMio6x6$h}>nLyM4u0xNk`0l4&?XYNonL+Hna={K{0(nS03CE{}F|?u=%= z(b|?h)a@WZnh02&S-M*4sg3}IuWyihW^!z_Ut}#SvIyBl4;v32OJ=0K;h|2nem7q{ zi6o!_Wz|;MtHE@l0O3#g%x@beda4Kpg2`gP&tDr5sBb8^8QStKk}!V4*DjAGpNatr z>pJLTS9VyunB7K~U{%5QbV`^^n*E`|k>g8oO81=fSHIDz(@w&-UM1$pPD@uR5?tKF zn;6U0%hQ4aF0p<7u8un?msCR&$KzQ7+y0y93Co3j3m`CcK5Tyb5&--8X{~snTJ{&_ zBEqswE=Ijwpiy%(R|~UZxjAT<4ifBUGaY`=)kOxeM;_xLydse~;ZSuK>=jw|`8Bm(H28c?j(#)smuOQ$ z&8~9}wd;~Oj@M*xx$S+BbIg~iH``Uf!qie*{K&qL-I&ciF;ZU0@KK#Xv(0S13KGhW z8$yYl?(iuTmXmMB-vyO*rn82BKTZ*J0!`WAR}$9+ zR({dFRtufKRd*N2S)1(nDj=WccmRv%!#Rs!JF=-_u_|LH35dvJN zI8#3makQE)l%Bav3-5y{Qr92}F zl_4SxclKn{o8j6|u8+6e{xTxaZlZO3lWl_6wXCf!6Sr@DN?fVjmBeAYR9VQoXPS9g zkQ{rfY4))zfWWab-*XJwn`yurgck;1UmndLJ+O?5`{ArH@%{drGOyh+a=Ed>w5%|Icfc~trvVc2s3q)ouuRLnn8Zi}sl_er$Q#?^= zwaUPZKU335K)d~+Iy5ICVPW12AU99?|GcUB|6lV&nLqul{t^`>6_ZpTy;Rra&o!}< M91QwN=6Udc0K2Me2><{9 diff --git a/content/ko/case-studies/crowdfire/index.html b/content/ko/case-studies/crowdfire/index.html deleted file mode 100644 index 98caae2830..0000000000 --- a/content/ko/case-studies/crowdfire/index.html +++ /dev/null @@ -1,85 +0,0 @@ ---- -title: Crowdfire Case Study -case_study_styles: true -cid: caseStudies - -new_case_study_styles: true -heading_background: /images/case-studies/crowdfire/banner1.jpg -heading_title_logo: /images/crowdfire_logo.png -subheading: > - How to Keep Iterating a Fast-Growing App With a Cloud-Native Approach -case_study_details: - - Company: Crowdfire - - Location: Mumbai, India - - Industry: Social Media Software ---- - -

Challenge

- -

Crowdfire helps content creators create their content anywhere on the Internet and publish it everywhere else in the right format. Since its launch in 2010, it has grown to 16 million users. The product began as a monolith app running on Google App Engine, and in 2015, the company began a transformation to microservices running on Amazon Web Services Elastic Beanstalk. "It was okay for our use cases initially, but as the number of services, development teams and scale increased, the deploy times, self-healing capabilities and resource utilization started to become problems for us," says Software Engineer Amanpreet Singh, who leads the infrastructure team for Crowdfire.

- -

Solution

- -

"We realized that we needed a more cloud-native approach to deal with these issues," says Singh. The team decided to implement a custom setup of Kubernetes based on Terraform and Ansible.

- -

Impact

- -

"Kubernetes has helped us reduce the deployment time from 15 minutes to less than a minute," says Singh. "Due to Kubernetes's self-healing nature, the operations team doesn't need to do any manual intervention in case of a node or pod failure." Plus, he says, "Dev-Prod parity has improved since developers can experiment with options in dev/staging clusters, and when it's finalized, they just commit the config changes in the respective code repositories. These changes automatically get replicated on the production cluster via CI/CD pipelines."

- -{{< case-studies/quote author="Amanpreet Singh, Software Engineer at Crowdfire" >}} -"In the 15 months that we've been using Kubernetes, it has been amazing for us. It enabled us to iterate quickly, increase development speed, and continuously deliver new features and bug fixes to our users, while keeping our operational costs and infrastructure management overhead under control." -{{< /case-studies/quote >}} - -{{< case-studies/lead >}} -"If you build it, they will come." -{{< /case-studies/lead >}} - -

For most content creators, only half of that movie quote may ring true. Sure, platforms like Wordpress, YouTube and Shopify have made it simple for almost anyone to start publishing new content online, but attracting an audience isn't as easy. Crowdfire "helps users publish their content to all possible places where their audience exists," says Amanpreet Singh, a Software Engineer at the company based in Mumbai, India. Crowdfire has gained more than 16 million users—from bloggers and artists to makers and small businesses—since its launch in 2010.

- -

With that kind of growth—and a high demand from users for new features and continuous improvements—the Crowdfire team struggled to keep up behind the scenes. In 2015, they moved their monolith Java application to Amazon Web Services Elastic Beanstalk and started breaking it down into microservices.

- -

It was a good first step, but the team soon realized they needed to go further down the cloud-native path, which would lead them to Kubernetes. "It was okay for our use cases initially, but as the number of services and development teams increased and we scaled further, deploy times, self-healing capabilities and resource utilization started to become problematic," says Singh, who leads the infrastructure team at Crowdfire. "We realized that we needed a more cloud-native approach to deal with these issues."

- -

As he looked around for solutions, Singh had a checklist of what Crowdfire needed. "We wanted to keep some things separate so they could be shipped independent of other things; this would help remove blockers and let different teams work at their own pace," he says. "We also make a lot of data-driven decisions, so shipping a feature and its iterations quickly was a must."

- -

Kubernetes checked all the boxes and then some. "One of the best things was the built-in service discovery," he says. "When you have a bunch of microservices that need to call each other, having internal DNS readily available and service IPs and ports automatically set as environment variables help a lot." Plus, he adds, "Kubernetes's opinionated approach made it easier to get started."

- -{{< case-studies/quote image="/images/case-studies/crowdfire/banner3.jpg" >}} -"We realized that we needed a more cloud-native approach to deal with these issues," says Singh. The team decided to implement a custom setup of Kubernetes based on Terraform and Ansible." -{{< /case-studies/quote >}} - -

There was another compelling business reason for the cloud-native approach. "In today's world of ever-changing business requirements, using cloud native technology provides a variety of options to choose from—even the ability to run services in a hybrid cloud environment," says Singh. "Businesses can keep services in a region closest to the users, and thus benefit from high-availability and resiliency."

- -

So in February 2016, Singh set up a test Kubernetes cluster using the kube-up scripts provided. "I explored the features and was able to deploy an application pretty easily," he says. "However, it seemed like a black box since I didn't understand the components completely, and had no idea what the kube-up script did under the hood. So when it broke, it was hard to find the issue and fix it."

- -

To get a better understanding, Singh dove into the internals of Kubernetes, reading the docs and even some of the code. And he looked to the Kubernetes community for more insight. "I used to stay up a little late every night (a lot of users were active only when it's night here in India) and would try to answer questions on the Kubernetes community Slack from users who were getting started," he says. "I would also follow other conversations closely. I must admit I was able to avoid a lot of issues in our setup because I knew others had faced the same issues."

- -

Based on the knowledge he gained, Singh decided to implement a custom setup of Kubernetes based on Terraform and Ansible. "I wrote Terraform to launch Kubernetes master and nodes (Auto Scaling Groups) and an Ansible playbook to install the required components," he says. (The company recently switched to using prebaked AMIs to make the node bringup faster, and is planning to change its networking layer.)

- -{{< case-studies/quote image="/images/case-studies/crowdfire/banner4.jpg" >}} -"Kubernetes helped us reduce the deployment time from 15 minutes to less than a minute. Due to Kubernetes's self-healing nature, the operations team doesn't need to do any manual intervention in case of a node or pod failure." -{{< /case-studies/quote >}} - -

First, the team migrated a few staging services from Elastic Beanstalk to the new Kubernetes staging cluster, and then set up a production cluster a month later to deploy some services. The results were convincing. "By the end of March 2016, we established that all the new services must be deployed on Kubernetes," says Singh. "Kubernetes helped us reduce the deployment time from 15 minutes to less than a minute. Due to Kubernetes's self-healing nature, the operations team doesn't need to do any manual intervention in case of a node or pod failure." On top of that, he says, "Dev-Prod parity has improved since developers can experiment with options in dev/staging clusters, and when it's finalized, they just commit the config changes in the respective code repositories. These changes automatically get replicated on the production cluster via CI/CD pipelines. This brings more visibility into the changes being made, and keeping an audit trail."

- -

Over the next six months, the team worked on migrating all the services from Elastic Beanstalk to Kubernetes, except for the few that were deprecated and would soon be terminated anyway. The services were moved one at a time, and their performance was monitored for two to three days each. Today, "We're completely migrated and we run all new services on Kubernetes," says Singh.

- -

The impact has been considerable: With Kubernetes, the company has experienced a 90% cost savings on Elastic Load Balancer, which is now only used for their public, user-facing services. Their EC2 operating expenses have been decreased by as much as 50%.

- -

All 30 engineers at Crowdfire were onboarded at once. "I gave an internal talk where I shared the basic components and demoed the usage of kubectl," says Singh. "Everyone was excited and happy about using Kubernetes. Developers have more control and visibility into their applications running in production now. Most of all, they're happy with the low deploy times and self-healing services."

- -

And they're much more productive, too. "Where we used to do about 5 deployments per day," says Singh, "now we're doing 30+ production and 50+ staging deployments almost every day."

- -{{< case-studies/quote >}} -The impact has been considerable: With Kubernetes, the company has experienced a 90% cost savings on Elastic Load Balancer, which is now only used for their public, user-facing services. Their EC2 operating expenses have been decreased by as much as 50%. -{{< /case-studies/quote >}} - -

Singh notes that almost all of the engineers interact with the staging cluster on a daily basis, and that has created a cultural change at Crowdfire. "Developers are more aware of the cloud infrastructure now," he says. "They've started following cloud best practices like better health checks, structured logs to stdout [standard output], and config via files or environment variables."

- -

With Crowdfire's commitment to Kubernetes, Singh is looking to expand the company's cloud-native stack. The team already uses Prometheus for monitoring, and he says he is evaluating Linkerd and Envoy Proxy as a way to "get more metrics about request latencies and failures, and handle them better." Other CNCF projects, including OpenTracing and gRPC are also on his radar.

- -

Singh has found that the cloud-native community is growing in India, too, particularly in Bangalore. "A lot of startups and new companies are starting to run their infrastructure on Kubernetes," he says.

- -

And when people ask him about Crowdfire's experience, he has this advice to offer: "Kubernetes is a great piece of technology, but it might not be right for you, especially if you have just one or two services or your app isn't easy to run in a containerized environment," he says. "Assess your situation and the value that Kubernetes provides before going all in. If you do decide to use Kubernetes, make sure you understand the components that run under the hood and what role they play in smoothly running the cluster. Another thing to consider is if your apps are 'Kubernetes-ready,' meaning if they have proper health checks and handle termination signals to shut down gracefully."

- -

And if your company fits that profile, go for it. Crowdfire clearly did—and is now reaping the benefits. "In the 15 months that we've been using Kubernetes, it has been amazing for us," says Singh. "It enabled us to iterate quickly, increase development speed and continuously deliver new features and bug fixes to our users, while keeping our operational costs and infrastructure management overhead under control."

diff --git a/content/ko/case-studies/golfnow/golfnow_featured.png b/content/ko/case-studies/golfnow/golfnow_featured.png deleted file mode 100644 index 0b99ac3b8f8f81dd990436b54c08f335dbf40996..0000000000000000000000000000000000000000 GIT binary patch literal 0 HcmV?d00001 literal 22915 zcmeI4cT`hZ+qZ)V3W6vd>0m=j5<&}Asvy0C(n1IjDWQZe2vU?H9q9rhAV?MI9i%EC zD!oaUE>fj_!Ewfn;|$OG*83+}3vy@g-(Ajib~)s%gTIQ>4PwFzga815SWZ?-4fEcL z`H4A=kNK@c-mZ;#Be0RxwFdyskRATu0OAs;0RZSTq=pV!M^Ql#jhRhT4Eb_gb3 zE$LIt>aIGDIWU_KxS3Iy|Tg1~|xs2~W;^y9@MN{F$9?MzGs)ud#8 zq{Dm?VKGOeZ3KZpCnqN^CkPkH&I|~KLZLtqH;|i~6Qkg?cd*bd_9zECI07SQ{e2*5M@)4;j`n{DB<+Z>M*lVx=A8X^px@;; zf&b-W<6vj`9kdA?h_FOhVe;5x7=V8Zz{VVfM%kOA{w3vK-QOYpE-@Nu`Y!+vd;Zl8 z;f(wjx5J)4+`f;gKTfqU28yDhpt2njGyGweQYg5?;c>}HiDEjqOpqpm5>j9gNE$39 z&dn{!CjbTuaC3{xKp@<_P>3W{K%7Sy_^%;;j;J4LeiL3sfR7i0rNF!;~#Sg!vFD`#zwhFQZAKdMGzss{7$7{j1&UQUpJfB+{C2m#@Q zflXnYa5%^W#0vqNfDnIGj^+N3&{7CHq$9#a#tvomeX`iV?CcST16PFQ&yoAr?D|3U z`_LCe{<9_?ZXQVxAB2aOM*wnor~*8a0+J9u2(J`KMjR@|FAV&L>(6>H!G-_mHu>i{ z`VpD;sK{8ZKO_G|_~)VibAZSq?a?SZ7p#uMf)1Atb%gETTmS5^ME;M&{^Ew2>4Gr$ z;ld9h-2!i+p zc|n+uf*{bJrG8iadk}k+DcT8UhY&ZzOfu0wwg#0yR|+Oc$-}LHpXsV1434xuTt3+T zjQe-Te@_56hgq8;OhkcyI{w-4tBWx3@Z?z{|9a`}(8U-deN3zwzCcRfEh2=fABR`|mq{M{A1`&U=p z|8~Xh{?+xTr=~R$Ez12P+b_*nUWeP&A28H0muLu4K3-l3uQ2e}=3hNjkj@B89VsMc z7qdTHNH7!Xhtsb;zxCAp=?Uh=6!|^fcTcRIAD(|q7Ocyu?{n#oR?Nj3<~kAh*M;KW zr`ccY*Z;Bc^N{|J17n9fPJ&J2Nc$Mqk%+Jb$GEU*9BCipIua4K;20M+jU(-2Tt_0p z798Wkrg5ZwjO$25*n(qR*ffr`k8vG|2wQNB3!BD~_A#y_5n&6CabeRq(muv@BqD6V zF)nNxN7~1@jzoklIL3ud<4F4$*O7>@1;@CsX&h-E<2n)%w%`~SHjN|gV_Zif!WJCk z!lrSgeT?f!MA(93T-Y>@w2yHei3nS8j0>B_k@hjJBN1T>j&WhrIMP1GbtEEe!7(mu z8b{j4xQ;}GEjY%7P2))W7}t@Aum#7suxT7=ALBX_5w_qM7dDL}?PFX=BEl9N_902M0Suu%p8VAPe964!7ZSV(Yj*6c1jsIgdd--v~DGDk4uLZqf!E~>wvEO00t zn6t9Tkz9Ok{GuqQ=V_7Fiz3&eoL;R3m1vA9xB9!|jd9G>F80^77RSGVz)&5zO3V&K;%R_rX zzeXW@ijMG8gV!ZoZ||;GH296?XPxD>^b3-j$>%xK`&HRta1i%Ov5Q1boxqK(!JI}V z!C~7XybDe~L7pjgQL*pOwRU7nG@vrgh5Cv5GM56*`%^ylZL&-sSjF?M9IU=LL* zJyn^8Txq14DtXHmikI1yN-en?yHHZbUb;7>81)1(&u%Z&z;Z?nI zjQC8`vovVV+;cDf7~e<)ir5)Ou#F*384%r8C|1kJuf% zADuKg=jktc3M`;z6YQg;JpMo$F1odn-O5-!wbEpIog1mTFd~gtLLV2ycR&i*P+vr} zcWL+JLbu!W0FMAe(1KKwA+3nu=(tCWi$!@0O@f zd7Wa|>h^ssDcirbmgRjLWfL$%WvjA^MhHz zA#Z|4BPi!vXl@0++gPs7FW9K#cTyg)&7OFg%m!yVDOsbWca7cyf?VJC8#2>oR9N`( z;AX&TL~Olor#Z#;hY6Fr()B@)^D*Ad_gtK^lwtOJGwm4 z=;bThXrM3a<#Vr)Dfx@vq*q7!bCDxA_HZ)1!F`HIDMriwxWR?ILBY$JTanELg<}m9 z>-sld;*Fe%WB(asP z&1Lj1fAJ04!`<8&m~MEHE>fBG{8c$3V3Mw$2f)TcE4dMfd`2Exe{v#W&waH3n!4=r zLCfXoeCiUv4n@cMK7FN>3jbS(Nb}3N$C5Dh^hE2St;y2d`k9hf#d{9Zd5j6`bNESO z118zayf_;PS7TL|p||X^hIKS6sGWDcgB@Y3EhvLw-2po88By1XS3~m|)5betrV9cC zK@W1{AI4J#)wU%z+pZ?vwCT^&PHv!T8rBZ#t}ZnCnDBJxvCxKAQOx@+~w!0no0XKhXo5#hDQfBmqRLbi#KKt1oN-YF-sFt z_J4Y{v*?iof}LPH!)VAqVIE<|3gF^={LpCc!V|lu-kVfvvTS~L;xnjWUY8782k~l= zEk-K6n$SL^@aBs0*Gtc4vL*1-iHRIS_A;MQt`KyvodpSK1VcA#e6{&C$AR;0D-qho zV`C_ai5?eU&>pOG*imou<1L~J2bwkup)c<@XD-%MWjOX!#LsErUVrMo1q!4*dsB+# zHjMCBlLS7N9KV= zJozDIH2%%Jt>J;$Ag@*&D+*HMj+V&x(q!CK4eCRWjrzzLqms^{mYyr+S*?Z`C5RIe z;QKX_LHLN2{mY-3tw!@{WV;|O)`gL6OR$U#^`6jADsg$S37Vp2QUNxDoHyY&YPXC~ ziR=~3)03`LVEZIiR5VnzCyVI;sU0ax`b@qWISDeAPL z`(27^un1M>h#(pJre=Gd!1QQL^kV5nNEK|f3*Mi431ak_G^;7~w)c94dIEIMj=E+k zv&YxNcJwjRvy*V5C@aQOq=|d#S0q2LE9!TD3lr7%1mmd%Tn}IuT0Fz*&t}q)7E+mn zkFV*-Plu1+dU5)Z!tn0oqX;;CoM6r}pEPu3tFAtbzIpI&y!GB2QzvQlULCeHvsqp$ zv$Y`4_9ve<(q3$ILKM%>`q9*uo&Y$bZ4Fw-Gw~WUJ6wI+P>`gx=UF@^7`(L zqvBtrpK0(UANG3k^iJHJeWzYj!r%!6VzGkr;tr$BwmaDTvCZ&khGM~S8L z0R^QKW$RlrHz=Iv&7#Nh*f%~Iy4?YD^!Vpqr*;gy|Jj6pK$uy0b2Fz!7CmNW$XT!d z=B&Z)u2b)&(f!q!Fr{cG{n};ENz(drS5YJ9m25YZSxI=TtmrLcqSTDdCdx~ZG=V|Y zR)zhBw{e*=GN?>KW$mWvim&$60~Jcw-}|+>i|d=TjGSQw$a0k3U5lGRJrNyF$z;X1 z(jx-Lu5^q?ZFY3-eNa@_$}4pn;ZmFu(H2HGe3(x>E!Sc#UQ10qG41Bm*^gddtM9*N z)_@<^cqwzXw0@DfYF1RoCc!%# zs#5lpWANZbE@X~?f~@>JG;G4EjuT~V5d(HGp*zd(R#1_J2lcmp%MeBgXpysl$#`&w zr&^Y7ZntiAmiaz>nfV2eSQb#1@4sbEyU@Sl?6@wz<={1X5AGu(iuk%$6Z=io|LLnp z8{~t~6MUMDBkAuyFVT*?+1Mr-uj@C$0nLd}G!Un3p6LoMUhLYLdJ#&Us5f+?cH1}f zHAs5>y;Ixe!U#eVNO2E3Vi+u7AKAD)`6PxC)YOluZ&XI0c+Ysw4) zSFY^Q#ILG94eT_@=%t$J+}{4Uz!{JQAPtqMkMv}(b19V8B>zCZZRS9a{;>Ntmkmzj z$D=y(%Iv0h$(_ASeYnj))fYTVfM-&rUe;?t#f7ZJ*Nh{5{;7)kBFroR|3S(}jZuva z%s$td%#Bcrv>P8^)gjgE%FoQ!g?de%ymDu7PiHG|>I=TZ?$i_2(ZGsb(;FGUH6J5E z>WR~5$hs4E?0w1CucQ)A^sRJw<9R!sAW~zQ^C{VO9z7WDF69bdzidXNb~8vJCi~L~ z+mTECdR=;TwW(=AuRkjkNyNAgelSdm+eB{c44ryV%7dtZ=m=bM=q-KQM`Ko)WayD& zZc_QiUvmHRl-=nMVn2TX6R!=Xrxm~ zMq5JPFq{`@d?!VpMr<#O7RMoexTouy4_atZd11GWs*9@bf?OqqNTN9ib=a$S=;TZ>}Wc5mj=4U)jR2!2KB z$~S(!C3Mmt%i5`TOg_6(`^i`7ka%o5#Ra7547*g0T1|%QeF?Z%^tjv^eQ+3PXjJ+O zntKa}mj+0>$nLbt;#z7$QFClt_V^1H12OourvUmX#S=j*rJ`)2;fc`htfx~3SLHg~ z#;fykA?5tFrbO?FCj7>Q2WN|g?CJ|)>t2uZ->^J;ukX(5E0ypjKwp#dPD^N^t`naB z?Zk6lxqctbZ%>!eO-xbElrp-OoKOqsLxPic`GZ8~#JV z>?dHuYTrIS-F*FW$|6vJpDB${{WPd4^qkeawyB1|jqDHjvi|K+oJQ8PM8gY-dEqZ- z62sud&xuqWiFtGAjfNhR02Xr=LWf~*34JMSo?k`v-${xTWCTd;#PO$XPNv5_P3HXM zmI@&siYJTs@`_o(6Z{H~_G{HxE-qKQ$h7keGb23c%~H6q87Vy6j9B4)$mpR|RoM`CRdM)C zlYYkA7%@eHN>-IG5r{!TB}f)*>0x$wUAPnTMLH$UywnkG_J%Eeg@6fLeST)2lQVBu z)cPf#EE`m=kPXU(ay>wIowHRx|*NgfaqaV+cD= z%*cmAILSiV75cz@Z=e-d6vA;I&u$G8Pp$7bIq%!eZ#M^d+UX%GD=u=VPK*krb@%3H z-YCX_2FaE=P(*~@91UX!w*@=6&94lWym>cC008W7^;iq9ysOx(L)%0`8be7puoBWeHW9BPc^lDG%^@3IT zeL}n&JqtP&Nn0;Isf4LLUmj|@(!j_3EK(n^WfkG!=EKgxE%KTL@x@Gj`6iC1!0ou! z@(B-S^3nwTrKTAwS$%CG0jjF#59yh*4-3Znc~5;w>WsR6+J8R4-gd?vbE#0$S)Zg$ zUn|7EF+Sma<;$b~Z(D=P(-R{v0>AV*1=xNoCAwIVU25%!GoHxj>@BF#qyW$xdmM3I zZ}JqLAYps0$cPTV1 zQai;?e#5tte@0}e&1O&-Gh}4-(RuC0ZFSqhk6X*U^W|-BC){bRJL3!E>x%iP$NGYcJ+#rK0PRraq4TWALsx!ym|2lK%@k#(h{?sk!A z&*Kw4RRHmuctNyffKU0`8@}}yII{ZbS7ZgODAD-nyDC}_*pqa88_5b+*5%KH(#(5$ ze9)*Yw_~i{>}Ju^9})|%vageMzK@T&sE9e3rw~9_;=&yZ01-CPDpY>4D%2+cECiQHiWqS4i zB40P%P(&OZ+}0sB2)_$!7d`XAGE^LO&@~SsN~;zGuYN%tz{j&wp1!v)qZ&=zIBi%y zg^vbC-zO(kjv4{i^R#5&g#>VA5a z<-9pD;f!+Q)r$)-kEY;n=c>Kmk5_tEt;2JSF5lwXLL4DZa_$F`K6=ZzlcD9`rEt6`buh^4Q&c&({%iS^D+4bh{XXRr>RM zAp|t_dwI)@%+D+Z<0mW@377N?dU)hM*R8rj`!eTEXFlpeIhJe--{rJ*CFs zz6eL3tlzWPssbAAdAVwQbYNoc@w#oA(Z`b1s&4=EhW2_qfc&L)=+YK<EN-;>;B}wMZ*tk6UOO=cMX~IA<92qt5fa4ud>F^*h{M zFVC7Df8r`KLFmJCVXY$<0qDe6;Bl#!+dQ!EEr1MXu$e3t9Tf4Zax>_ja|kfyc($vd zTV?$uNyU0Znm@S9*YYDd#Y^S!L4MXRcta{2Ie_H-@iyD$i#++_U-KArF{`{q36m%% zp#;p_d=+**ylZDw@?Z!@(!to|{!1Y1NY4&>Z0loxqGYf5#qq`28yb!8o%+u)M)6dL zPqZaNO6jhryRE@GoCm0sNCCT@qjA&l1g~}7ya|%{*+NDGwq53lT4XU5V`O(u^?aJu zYI}0OJL+Q0BuLWhO5{mOjqEuaq>Mx$|IHVd54Lk&a!Ry4R$t_K9v{^!raqpNi)!@* zxi(}krRuk#qJ{Hl8CrlSTcuU3O0DrtUEh-JD_5#8qHurLDG z0`s}J;^NLz2`b8#Np<@YT*ha+CF|#pEEHOS3gbG6i`wW^+HVs_6!sGD9$Z!iq26{r zHmfSFHg?cf0pgzf{NYW~bMrOJ(xuq&kmX&s*}E{9NbFUCnTNL5nzIg^KQ#0z(0&t$ zP@64tI#DanNFTc~HJY=sn1dXo=6elJ&@&+2co6ipw^%Dm!QDpsYjg7eAuEYx{zJMT z@zWEXY5~3Ye8~-u7%<|M_rs6xUcM5|7D|q|0oMlbi*4YfkM7|I^SI%FoF(xYT+H}O8v@)Y zr}GpaIVdewSY>k}1x_gu(G`>@G1vNhbV<@_5B*H%7)C?6nANr;V#;*MJCx2>PTKnJ zS-t!jIfn9d^SLn7<>H*E_J|#s5eqc+R_ zygLE*gFH^e1?AX#Jf~~m#u7QJt=Mi*?v+p z<<<+6q^6UN3jKP4i67RL$cB|gt`quiy)KrCj*H~HRB2VM;u0h&ujQvRwTFU+uMM>z zN;==xTogmjw+ zBrjTdZ@*dXskcK2;b)h%XgN?*IOYZ-R{H4OHof}z>>J3U%q_qvuQ4u#ZJiM>OK4OM4MzjNi$ID5?;sm&!6?v z@@Ud)M0%ylh~HUzOR>qJL-J6mNfXp@%KdG$!vlkn#jFwO-`s9?UD<37qtUlffXyWnIIC_b7{0Sep6N1RqoL~}W%<<4yOGS8lx=`kw51lBN-O;KIlLdA?^Rw7Skeiz1v>G8DH@UdU5}e)4c=ty) z=CX3`l`l0^0BLzQHLRX9EsdMkA(GCrs8Po}UWwXBDQ*p64+uPosNC5*;TRwMHMJ~$ zutLgI;+8yZ4DWnF$>v^8c6re@W-m~FxVh)cI3-~yD&zEo zDJ4X~Y4MirX65=DYv2=O%_#BcJ|P~5*J4Wv(3MAn8QHX9<3q%`)CxsJF)7i*JDRmT zoIGN{J0GT#I!0H%6iFBEc;xlq1~!{Dm1o~5^YsRw#Y~AMp1dVq7AjrxHzmb2rsW(+mocDzOxLeaW{hDG887MUUz~2DzozpgnD0NqO`pa|;Pr zzMJ}FNI#XI#dcJTEvs7Pgac1!DP3#>$z1F2&KS9m=+Y zAHkA~OqLI`GDQ3Ct%WQv^OyO$b&lVr*YGUEnZ?mopiZB-vHJBRqk$^Vx@C3A89=$j z;^4CWew3@cM(w+WtcKCxI5SFARh(C%otJqZ5rdw)iEcw&7L~h%pQN6urTpcZ3sDym zA*)2!>=oA+_uKZ0=&J`0+}?QLPwRU0ra>NDxsle#adW>6ovpd(CU)1cIiy0NKePSq zZu)v#z0hTnY)j&29qPJYAG{YDQ0PKN-;tjnvuDrqGD&HUW*k}H9iTsDZ2+=cQ_iC{ zMCTgnZJ{m^oYT!Kn*HM6WOd_qw`s`exTE=W46I-b%BnWZ2Hwb=DIXo6?3^Z^Qmhs! zO1z~F%Dz}eFQe+1Em{b=#5R!fNN9ND@LSM0t-M~Odgt@zIbAtxLN)<-kiFmamUFXU zM{nNEQ%;NIRIHA}g^f|1 z4ZIy#Vy1+jyd*ZF`*PDcP?V2-gqvH$bmM@El|X#g@If^w{@-o|V{XU(|J|Scj~m69 rJI?>n`|r1r|D%wfWeNaymjRznm^fOf&xIcTn`Aj@C8=Bqqx=64y0Nx8 diff --git a/content/ko/case-studies/golfnow/golfnow_logo.png b/content/ko/case-studies/golfnow/golfnow_logo.png deleted file mode 100644 index dbeb127b02a270963c61bb29f21dd9f2d9b576c9..0000000000000000000000000000000000000000 GIT binary patch literal 0 HcmV?d00001 literal 8858 zcmd6NWmFu&wlxF-1a}DT794_0fZ!Tjf($S?%-|4w2yVd%8iF&p2A2T_hoHfOYjD?( z+;{JN-;a0S@7JrVPgSk6_c?WHudcOvb%chRJQg|`IsyU$mZHLY&1XCN+yc;EJU`Dg zIswn-wVSNIn-<8*%@g1PM3AxsnFDDQ9Rb!rO(4M1+j#&ehJb*?Vymt1rmw0ZVgYjG z1pM{k^l}70qY)6qB)z}@3wxj&jXBWT)=8Z1sHKCB#@14tPVcQMw<=f$Xk)AJ$pxtO zNln}0lf8woC7q-MjhL6nGl3(}4M5}N=-}ik;w4V^4_=Yy{@-aXI+}mDxY>)-{Z~@@ zsv0yhAQvFbTTTHE3m!fp8bM)Bp0~n6y!`AmyxcqjT-?H3Jp3HoJR;n}BHTPQ|Gem) zqq$gGiDQ;0XRVt&{6N%k*3@E-wI>i-(i@ z?~?xIsH*zEyE;1l+uPMm6Zqfx{*T12+TLIwmnP5^73_&8$w>6E-XCneKXKR4BOEojSV<^mk1-L0VCvK*^S)Ju|_ms&oAlDxwntK+-k>7sv4b{Fv zAL>4UwpWKyNxGIge_P%oJDp&U<3Wt2JK+y5oP7xDFzLlf|2qYP|;j=H?j-IFs4OFk+qJ}wDl5~|0jU>&MTThR~l zZGIj*)o+xowhX%Y0>7S@;-yLs<9Lj}Ds84vj!IUq397xhyv z%wMjR{^cW)?rpw5S7fsCG#4?mcK-SFpeKQS@ehA{<&rM6ob1q5LMW_C*<`-@WQxAS__95LLIzrAFqUWsx!_sD zMZ;VVq~j#^d((2PpXfeaE80lmh^6!@#bW38d^o;zyA^=RxjPk3c3bJ^G&|HXdiTO% z>e1qt=#&5ND8-dq7GdrnzZxRVMDhH)Dfii!xdH-r_&ITMnW>l}BrmH4&b~NjicwtUfxCq? zHckSU%^w)*=7P2awC~zj{>dXz5tt`!Bi`GOG8K3Pe`t@Slg9iucTsr=+N0=`jv~Qh z!DsGvmlHztj(KHACibA{-XVJYOGgWh0s|LJWc&5J8x4Vpmkwv7zrKtiQ~HHaSv>HC zCI!>gnGPdls@EoT3neT`7{Yq`ClWofKNb|BRIWy&H0 zY?3<5nVUT}I90N;*~;Wl99pPL$Y6ZeK+X6-~ZyX9zXgh zb#=QbTa}GJltjGn8sS<4N4U(-%YoQ~@9lDYpu_yvG> z>8}MV_u4XTo$;9vHqL9PR8hY&pQXV$IzEoei~e7%gOY~1?KPE93fhD&>L#bd7jm|7`x4- z%2*ueR8&#EugiBob;p(X!pEm&t*70RhM}Uu%8bOb`=VsY$9?~zI7GWhEU;Z$x1IYO zW&4VQ*!AJ+*8cWJYrs*$q)axYypfsC%U47Q@7b;}?~rb#BZV}F)=vIsVTH(WuBZ!SL3rO;@kMANCRs8$=zsB1{NSSd zUH5!SBy16GShm2{6336($%{E6?PM_A?4vuH#O4d$FY&N>)odFi1_;46#!c3Ae#nfj z=c`qkh!!arpGFKjQVDb1FBuirF&-xi_QOrZQbXK#zu(ONG(vYpFss(K9z)-1`QBWt z558Ivvgx@)X=A85qJqY~xbvNTObdaXT?so11DAe*!SUUfUt7&2An5fU!M)9IxG}9J zON$RG^#WfdvTDDdxA~-YTmu~)nXgu1dMs$i-m0MH4-Z58RuacW#z(#%vdnl`rpGzq zgm#)G1vFt)OJ4n%Wl&$I2Uo!nA?=)=h8N3G!$~kxi z^oO4U`WpL5e0qN4lh)LH+`M z#8s8{Jv!eqcrVg+_t^CZ>(MICn{pUc3bxtz{mAt)pJVv|@#7ONuIUoAqSW;!Qo$Ed z@?;^)%`Tth(iQn~J-xUM&!Q5N*RMbASEJZFR&Qp6AjirJj4XjD_m5r;r=@xi;df1Z z+DJdmQ$->&h}iZCd)lvmXf)n}qe2rPq8qmc1E9rPx=@GuA7&$7+E#-AI-Ojnz)&0X4{Zf8ow6ZFS zfU8ZQFmcw0ynfXw#*F%xwsOUp416CkS$0*^>8G>!;kzt*?e=Fl+7~HB?0lp&9~5yv z3PTv?Lq~hh^MOB<+OrsFv9@cRfABJWo7}cbz)9-8mWI_m%V5l@5_`X^YFg>5NlUqd zTDC1+jSm8Tfyiz;w-BY8KY7W zL^n+@6)TB8#Gh#y?0!<1l#D0+RR@82)}Vbj$+ybuL>igafSVx#p}upGT^x^W(TQJU zRo!(N{7U*&?DVTQIh=RBMA8VL@l{6Ws>@d zJCxszsaDIJC8)6$sEZsDOf{3A;vISA$6NWGHoAjI4Ac$`{R9iKrdtQD%$`X4_lcT#J0lgd&@R?KEd~zBtRsU({FLe; zKw++t{_>DikE!LWnV*10$$@YF03TJ;92pDklh#FlqE@pu1J@1hJ9Po{@8`Mp=%@!3 zb0zkOhXmWv1SV$9lz85{lA0KOHBP+>2q4xB?RkOH(kq06a>XSqztbaU*y2sdB zHWD$wknjB~ANlNElz~^{))}ZhT}nZR)SRqSEa4G)H4Sw1qP;X%T>zb!rORW;3Bgr7 zV=w7G8r)iAkUoITrb+oidjE@6tC5{)6PTz8rK@7zcPf&=cf7VM7X|n$oR6iU#>Ma! z0%4#)HPv>XojPaXN2kFkLb)g>{k^SKgFhy4`Lvg1L9OT6;nPa}$SBH+vo7<*2euZ(mM!ZOkdJYALJJrZRLwvpShHB&Z6K}8&Pjv{WJD)g0JM(K;uNa| zh2F5!6jQUag*Rqj6tF4fF!c7xs$szW@$as^zV;}8@`h9>dc}MfiaMJbOjvl!L`3dU zO^XMWZ$Er(OIC>wG2^X+yme`?P@t3Qq<%i`NZ%@XGv@|*-PPa+8~(I&e@Bx2t%r2d z28y7@hA5dKQR(kzG(DRvjVlV*=sUq9myRUkrC1VnQKP~g;Z7>tJ7pw7e3~+kOAFQ2 z6NKGtejX7}$$3uyEWX8o$~%zkuvR%Tk?Vf2$XcYMxb_1DB{a~hy&8nPPLC)pUC^j; zc!2EQYN4-fa$1BW&*)5A<}=aE&%dBqwUv9Np5czZTlJo^sTVbD>}kWjwl-nP;unJ9 z%32-NC%rr-VxcYeU9l_slhm{;4!gJe@Jz0KLMhR~7Gi4L_H`8o+75zfHrDN7u2s1j zW2re$14-<9=oNLG)JTQ$)QjkSHw<*`-i=c2v=uJ0DX+NNSeTH_L_OB@&Fe;fmB{?9 zBIBi_y}C(bHyWBAT7;QxJE~Lt5wfAcmZ8;sf(3LGPabp@jA5+~S7H${{~Iz`p=;tf z3Cjwf(EV@8Sj2=i112VHI??c`p8V6qi$qcxtAzLsq#!VyA(fNqR1gr)6o5fdew3Ik zpR@Inh&sDiFWLzM8>^yiVf+4Zq9q$TzuDD&_XQxIw1hf;$v)7=ZlfgIIEs`vh^hLn zF=v`Znku93n~2fm!<2P$hlvqw$i5K@Mdd;9WHD>4$>!Cqb3!1c-fpt_X_&8W!3b;5 z{l^yr`1P!1;C6LnpW#nP6%y7y%@UOZ9~eC&bB}oW{G6j67Jn||^v(ZEc~t#^(u*XXSWySIplOvA&= zW&tWH**&CFgt3bi0*%1QqsOFN|!*@!LaH(^Cm(bH^>@J#1e}IS0yo=!a=SDe9tLPFh$0bN8kNGHI6WIt%RuH;@`E79Md~ z7QPr+6(L61`#Ii^SOMA8Wsq*oc^71<{1Rp6qQEr^T18x!1!bWpEbux}DEX(ecD55_ zAqV|HeNXK|Ap|KezeGj8yw@}8wLjV8W})Gce`2;B`}5|^2@HuY+JuEI;~PlDG(L(IFne>(1=$I$dd!6L_ry!=@%y z``+`a#vV)x>l-DxhZX6#1tvOdR7DaNrQqyy;TjkRy6aOutF>e44~8ty1}!;vy5Pxm z)4M_s3zufPuzt^vh76LTO`QU!Uq4-m!a88VC3fSV$qvuvPbyy@($?~-3UMZ8ssHlR zjbbvcWHP``Y{bag46RsgX0nIKIX{(_7*QFiet6&fp?=GA!Sk@hdbcxX-^%a$I{oG* zg*LTvaw|LD@o{qj{%yO}%HvWpuF=oAO3_05s~hCROt-M#8siYTzkBbIUTjpk+N^jk ziler7n@}e{`0(_F>b)JEzRMNhjWi#;_M!LTV|- z&!^2gGg)aoyV$i@5le5GmW$?h^r=g}`;j}!<1YiDb4G2sRj^9=wf4;^p@5rIlk1Fof%G zR&_dj?yi%hhw}EVluqt5zqLK~W##5I?Qt#;H&WMdkFg5{?Dk2&ExAHoTuo{6*IRf@ z;m}kAjN>8puGaoIt@GY?5G?>m>wWT>WC|N&iqW>vEOl&R({$z07ka~7?P8;^1PPPG z;nzoEm&dcPxs;!cZsF7AKbNTu-ez#etVy`YLK;P=xZChrqD<{&{v|2om94QyTFGZq z`=SYicbgoijvJ~|)VXUdI6SN{uBKLwn&?aK2ugc^N>X;b`7t3qrk^4@aqGOT5zJC4 z8%(ZdmD562Tb7wx^k9P)Cs7C)MAaitGT~}lS2%0%+ot8po%4eM3u=oBw)M{~4DfQr z$ES|yLfjbqrQ(IJIkJJ79wrJ!SkniVWok2n)wHfT-@%mq0wvN`&EM?2@nLj9u@&*I zA82(c-tebdB+B}FtWHO{`c=FW+BeoHGco;Lkmc^Ib9$(Eu_PnGo`8>1!`>Vc0+~bB zf3m+bTrd5z(NyS)#J+ykuW8?b6*yc$fFite4p)}gT_RoJH^|_5<6ce(C%&2ZQyRsF zdVPsU8|u6MogX`|TdPOD%tVSM!7#Be$$XsAW=do;#Sg3FcJi>Cc>%tl4HhdhDo1}A z`-b0sJ3lI*e*4(L^{*oK} z7IG03V*+t5=Uda#T{Fl{KCP_^o?c2}5tI;YZ6W6987$wBnc9e z&xK;3_-*Em+b5ku?~g$bm+@PLw-Uy7h7mIcX0HrHL$)II%VrZz~Gx$rS)4|9wH2|lwl!mQsfss=1cuH|-|7>91oNVfP`j+0}bV0_QS zxCWV+fSYH#B7+RFmt21MPv&k#xNSAWp!$f2K!nT9wC!lVFiP~1aP{eC9y+6sc>f^R`$J34qut?R3@X^`G zT}!5e^Fu9-5l@GZ7E{BCim@=~P<+a*%-W!(vD&a~5rtND!j$q{D9vY$ZPys0l|IaL zmA)rs!4cr=-Qi%=u6^nmz!;9n^3{;18ijqtra7?a;U|uIxevh8C@whYg*v7S8X|_7 z;-sY8bkYdiNPl&Y_)CHIDd@mS|6y%8F18V?={tLEgT(gPg1w$Z{kmnMjt2(o`yQW4 z7LGv$+_BRFQ3ej(?LZi6gy;->5^xrNRR!U4{<%$;D1l6IcBzZD?-Bs{z!2Vp#G_c< z6Z!yR*V-XK>@4sjr(8{d;B=Qn%npe?9G+}GH2|&M8}&O63s&Zgtrf=Yk~W4KIU`G2 zJ2iOaz>49%Rg7CVdC^#3?Ab{g#loAvK%6N|%vm1<*fUis+?^60=ctFhuY(i2Lx!vN z@Qhm>i$of74pme&r1^Tk{S@*GGHHRjcD{so7QzIhf^$70KWi^vi?43@8-@-R3kqk- zBm~f4S5?Cv`%*(h(7Y!()%Vr8OhWp6@aE9%qlT zdg%4Oe$PkevWzxYB%a(dvdS+=M=O7YEXRiBX*Q^l#9HMjP+b zDfg04(t=cy?AE$ogz*5kcs|BVbtc93tdB|ssKmM3eF5V7lBC9kWGE}Hk;o9E>s~Hq z1{fJ}GQJ$d;rMN`S9_<2xNF(SfMnx?0{2IF?Cag|H_Ma+5)=e_n@fm3viKW2SLP;$hH2NSM%zO! zpquJ+^sm8~BV-)S&tH^YPCsp2-pbjV*Q}Q3aB?V9)1wM6=@;i>dTlfW*jVflDe~Ap z&{?S)Y`7zDcq+=Se`6_6LztR@J%A|fyu4?dmg@bUU!b_29B)3kz}RsWA(5nSq@c-I z^5WN|wJ844D6kAqc6|wPVh_q3<4gz=N3Szf{#pmmsRtZAZtg~qWoo{FbW2}}y<-T%Pjpb??8;Z1Dy3TS-Pe7bF=

Co!VWdh$tGc8PQuMiXrEdE;5v=4p?3 zeU_fOtxdmk=upu%6TT9Mnx+&;ZB+PX<5X&0E?m2^tpENr7kGv~VQ$SJbvE5mxd>K7 z{5JJNnx03|+%vC?qMh$qZznnI&YCe`HEJUTbOz2EY%d%g7*hy|=AO0ppO+(l< z&?AQxhKh`>+BthU6W~>RgX78ZcppS`lY&ves*}$V6Eg#Lu$_RM42~(7qF4^G+0YJa zr8MPN&pKbIoaxms(NkT{&j~Y=tr3Lvtj9?{I)JS|oHHytx$x4V%E7?xV50*vkk+4FYfdndT;L+j{7)Xm()yqi4l%n;ak1$xVx zy+5txfLbhdD-29vkj4DU`P(dPGG?@&E?6zDUJ8U^RLA2Ic>YA$8Mm1NMW**C9%&k- zteHOG3;ud@=^4HtEo&d_8YP@_PQ*MK=SYU)$b*7ckYhkjw5H>h^%~A=;weWbhG5QYklHdGHQho~8CT}kcTb0dJg(nq8rC{fUm>(QY{&4QXc2bbC? zKNOoEi=)5M(rE1m=56IoC1XVKp2tDiUh!U4<3bxujT-JuD4TvU*lPK(c#y9=lz^_O zLQq-m+0%IS@v3XD0b6ic^5Ak3bE8JCykAY5xX>n5eW2Y<2|Xgt(W~ixc;!lP#&gcw zTw4uEQi&`y?%*CVr8DSD5lPY<`pi=z)Jkn6Ja3HtG562gBdohSqE9DjNJfe7o`&W# zFHKH(vI&-2<|(&ZGkmcO#~-w2oT9H+jaQ>3@<bA|N50A{82n^kg>w{aZv)R_%R-v>D|8 E0r1&*@&Et; diff --git a/content/ko/case-studies/golfnow/index.html b/content/ko/case-studies/golfnow/index.html deleted file mode 100644 index 9e82d90cd1..0000000000 --- a/content/ko/case-studies/golfnow/index.html +++ /dev/null @@ -1,89 +0,0 @@ ---- -title: GolfNow Case Study -case_study_styles: true -cid: caseStudies - -new_case_study_styles: true -heading_background: /images/case-studies/golfnow/banner1.jpg -heading_title_logo: /images/golfnow_logo.png -subheading: > - Saving Time and Money with Cloud Native Infrastructure -case_study_details: - - Company: GolfNow - - Location: Orlando, Florida - - Industry: Golf Industry Technology and Services Provider ---- - -

Challenge

- -

A member of the NBC Sports Group, GolfNow is the golf industry's technology and services leader, managing 10 different products, as well as the largest e-commerce tee time marketplace in the world. As its business began expanding rapidly and globally, GolfNow's monolithic application became problematic. "We kept growing our infrastructure vertically rather than horizontally, and the cost of doing business became problematic," says Sheriff Mohamed, GolfNow's Director, Architecture. "We wanted the ability to more easily expand globally."

- -

Solution

- -

Turning to microservices and containerization, GolfNow began moving its applications and databases from third-party services to its own clusters running on Docker and Kubernetes.

- -

Impact

- -

The results were immediate. While maintaining the same capacity—and beyond, during peak periods—GolfNow saw its infrastructure costs for the first application virtually cut in half.

- -{{< case-studies/quote author="SHERIFF MOHAMED, DIRECTOR, ARCHITECTURE AT GOLFNOW" >}} -"With our growth we obviously needed to expand our infrastructure, and we kept growing vertically rather than horizontally. We were basically wasting money and doubling the cost of our infrastructure." -{{< /case-studies/quote >}} - -{{< case-studies/lead >}} -It's not every day that you can say you've slashed an operating expense by half. -{{< /case-studies/lead >}} - -

But Sheriff Mohamed and Josh Chandler did just that when they helped lead their company, GolfNow, on a journey from a monolithic to a containerized, cloud native infrastructure managed by Kubernetes.

- -

A top-performing business within the NBC Sports Group, GolfNow is a technology and services company with the largest tee time marketplace in the world. GolfNow serves 5 million active golfers across 10 different products. In recent years, the business had grown so fast that the infrastructure supporting their giant monolithic application (written in C#.NET and backed by SQL Server database management system) could not keep up. "With our growth we obviously needed to expand our infrastructure, and we kept growing vertically rather than horizontally," says Sheriff, GolfNow's Director, Architecture. "Our costs were growing exponentially. And on top of that, we had to build a Disaster Recovery (DR) environment, which then meant we'd have to copy exactly what we had in our original data center to another data center that was just the standby. We were basically wasting money and doubling the cost of our infrastructure."

- -

In moving just the first of GolfNow's important applications—a booking engine for golf courses and B2B marketing platform—from third-party services to their own Kubernetes environment, "our bill went down drastically," says Sheriff.

- -

The path to those stellar results began in late 2014. In order to support GolfNow's global growth, the team decided that the company needed to have multiple data centers and the ability to quickly and easily re-route traffic as needed. "From there we knew that we needed to go in a direction of breaking things apart, microservices, and containerization," says Sheriff. "At the time we were trying to get away from C#.NET and SQL Server since it didn't run very well on Linux, where everything container was running smoothly."

- -

To that end, the team shifted to working with Node.js, the open-source, cross-platform JavaScript runtime environment for developing tools and applications, and MongoDB, the open-source database program. At the time, Docker, the platform for deploying applications in containers, was still new. But once the team began experimenting with it, Sheriff says, "we realized that was the way we wanted to go, especially since that's the way the industry is heading."

- -{{< case-studies/quote image="/images/case-studies/golfnow/banner3.jpg" >}} -"The team migrated the rest of the application into their Kubernetes cluster. And the impact was immediate: On top of cutting monthly costs by a large percentage, says Sheriff, 'Running at the same capacity and during our peak time, we were able to horizontally grow. Since we were using our VMs more efficiently with containers, we didn't have to pay extra money at all.'" -{{< /case-studies/quote >}} - -

GolfNow's dev team ran an "internal, low-key" proof of concept and were won over. "We really liked how easy it was to be able to pass containers around to each other and have them up and running in no time, exactly the way it was running on my machine," says Sheriff. "Because that is always the biggest gripe that Ops has with developers, right? 'It worked on my machine!' But then we started getting to the point of, 'How do we make sure that these things stay up and running?'"

- -

That led the team on a quest to find the right orchestration system for the company's needs. Sheriff says the first few options they tried were either too heavy or "didn't feel quite right." In late summer 2015, they discovered the just-released Kubernetes, which Sheriff immediately liked for its ease of use. "We did another proof of concept," he says, "and Kubernetes won because of the fact that the community backing was there, built on top of what Google had already done."

- -

But before they could go with Kubernetes, NBC, GolfNow's parent company, also asked them to comparison shop with another company. Sheriff and his team liked the competing company's platform user interface, but didn't like that its platform would not allow containers to run natively on Docker. With no clear decision in sight, Sheriff's VP at GolfNow, Steve McElwee, set up a three-month trial during which a GolfNow team (consisting of Sheriff and Josh, who's now Lead Architect, Open Platforms) would build out a Kubernetes environment, and a large NBC team would build out one with the other company's platform.

- -

"We spun up the cluster and we tried to get everything to run the way we wanted it to run," Sheriff says. "The biggest thing that we took away from it is that not only did we want our applications to run within Kubernetes and Docker, we also wanted our databases to run there. We literally wanted our entire infrastructure to run within Kubernetes."

- -

At the time there was nothing in the community to help them get Kafka and MongoDB clusters running within a Kubernetes and Docker environment, so Sheriff and Josh figured it out on their own, taking a full month to get it right. "Everything started rolling from there," Sheriff says. "We were able to get all our applications connected, and we finished our side of the proof of concept a month in advance. My VP was like, 'Alright, it's over. Kubernetes wins.'"

- -

The next step, beginning in January 2016, was getting everything working in production. The team focused first on one application that was already written in Node.js and MongoDB. A booking engine for golf courses and B2B marketing platform, the application was already going in the microservice direction but wasn't quite finished yet. At the time, it was running in Heroku Compose and other third-party services—resulting in a large monthly bill.

- -{{< case-studies/quote image="/images/case-studies/golfnow/banner4.jpg" >}} -"'The time I spent actually moving the applications was under 30 seconds! We can move data centers in just incredible amounts of time. If you haven't come from the Kubernetes world you wouldn't believe me.' Sheriff puts it in these terms: 'Before Kubernetes I wasn't sleeping at night, literally. I was woken up all the time, because things were down. After Kubernetes, I've been sleeping at night.'" -{{< /case-studies/quote >}} - -

"The goal was to take all of that out and put it within this new platform we've created with Kubernetes on Google Compute Engine (GCE)," says Sheriff. "So we ended up building piece by piece, in parallel, what was out in Heroku and Compose, in our Kubernetes cluster. Then, literally, just switched configs in the background. So in Heroku we had the app running hitting a Compose database. We'd take the config, change it and make it hit the database that was running in our cluster."

- -

Using this procedure, they were able to migrate piecemeal, without any downtime. The first migration was done during off hours, but to test the limits, the team migrated the second database in the middle of the day, when lots of users were running the application. "We did it," Sheriff says, "and again it was successful. Nobody noticed."

- -

After three weeks of monitoring to make sure everything was running stable, the team migrated the rest of the application into their Kubernetes cluster. And the impact was immediate: On top of cutting monthly costs by a large percentage, says Sheriff, "Running at the same capacity and during our peak time, we were able to horizontally grow. Since we were using our VMs more efficiently with containers, we didn't have to pay extra money at all."

- -

Not only were they saving money, but they were also saving time. "I had a meeting this morning about migrating some applications from one cluster to another," says Josh. "I spent about 2 hours explaining the process. The time I spent actually moving the applications was under 30 seconds! We can move data centers in just incredible amounts of time. If you haven't come from the Kubernetes world you wouldn't believe me." Sheriff puts it in these terms: "Before Kubernetes I wasn't sleeping at night, literally. I was woken up all the time, because things were down. After Kubernetes, I've been sleeping at night."

- -

A small percentage of the applications on GolfNow have been migrated over to the Kubernetes environment. "Our Core Team is rewriting a lot of the .NET applications into .NET Core [which is compatible with Linux and Docker] so that we can run them within containers," says Sheriff.

- -

Looking ahead, Sheriff and his team want to spend 2017 continuing to build a whole platform around Kubernetes with Drone, an open-source continuous delivery platform, to make it more developer-centric. "Now they're able to manage configuration, they're able to manage their deployments and things like that, making all these subteams that are now creating all these microservices, be self sufficient," he says. "So it can pull us away from applications and allow us to just make sure the cluster is running and healthy, and then actually migrate that over to our Ops team."

- -{{< case-studies/quote >}} -"Having gone from complete newbies to production-ready in three months, the GolfNow team is eager to encourage other companies to follow their lead. 'This is The Six Million Dollar Man of the cloud right now,' adds Josh. 'Just try it out, watch it happen. I feel like the proof is in the pudding when you look at these kinds of application stacks. They're faster, they're more resilient.'" -{{< /case-studies/quote >}} - -

And long-term, Sheriff has an even bigger goal for getting more people into the Kubernetes fold. "We're actually trying to make this platform generic enough so that any of our sister companies can use it if they wish," he says. "Most definitely I think it can be used as a model. I think the way we migrated into it, the way we built it out, are all ways that I think other companies can learn from, and should not be afraid of."

- -

The GolfNow team is also giving back to the Kubernetes community by open-sourcing a bot framework that Josh built. "We noticed that the dashboard user interface is actually moving a lot faster than when we started," says Sheriff. "However we realized what we needed was something that's more of a bot that really helps us administer Kubernetes as a whole through Slack." Josh explains: "With the Kubernetes-Slack integration, you can essentially hook into a cluster and the issue commands and edit configurations. We've tried to simplify the security configuration as much as possible. We hope this will be our major thank you to Kubernetes, for everything you've given us."

- -

Having gone from complete newbies to production-ready in three months, the GolfNow team is eager to encourage other companies to follow their lead. The lessons they've learned: "You've got to have buy-in from your boss," says Sheriff. "Another big deal is having two to three people dedicated to this type of endeavor. You can't have people who are half in, half out." And if you don't have buy-in from the get go, proving it out will get you there.

- -

"This is The Six Million Dollar Man of the cloud right now," adds Josh. "Just try it out, watch it happen. I feel like the proof is in the pudding when you look at these kinds of application stacks. They're faster, they're more resilient."

diff --git a/content/ko/case-studies/haufegroup/haufegroup_featured.png b/content/ko/case-studies/haufegroup/haufegroup_featured.png deleted file mode 100644 index 08b09ec9db8b7792d07a9a4d385cbd546d2446ed..0000000000000000000000000000000000000000 GIT binary patch literal 0 HcmV?d00001 literal 5841 zcmb_gXH-*b)<#hTR8VPGP>4Z#2`!=b-jOOT2qZv2l9+@dLGX%7vjs(Z4^0F_I-w}i zI|5QdKw78)3r*^Wcf2#-TI0;GIqNy+ea~L|+3#*Y&e@R`X8LEDxtZzc=*}7%=vW@D z?MHV!Bg4_tjJ@0LXt{{fy^ga&dEf$EF>pFf7|IO}F!Xixgj>R0VYhF!!&T_$=!Fs1 z*KyZPO_bbGzLKs#btD6Q(MN1LIu-Rmw5z)h90zcNdm@mkz%^0>5P*QG0`0)2Ql@Ba zxEI182m`kYGP8CM@^M#$0oBz2DuGHz27KW-S3sceO(a$+P!;%>U8SS_PcaAx_)7)n zqYC_oQ`b!`0NN-F8~~PtNVrSONde>)C8Z%!a$sq3fQ*#1JV*)*l7dJ`Nh?VyD9K0z zeqF$$(=ad(B}*N>UuPY?sRF%lIJ6Q76c7*~86Yc(!gzwD6%`dhQZgVJ8Hpne3G8ho z&NWa1iRJ&>f({((jzOSt2ow_V)1s>z$`7XsJo5A(A^4(AP5)^aiT#zRqm+RHUC|(E zNhy%8@6WjY(#GN};s0{ux7t|i+h{n*5{^arVcd`E;lckm`6ze){h*(UN8Tvupxpg@ z;YggJjwF2%EaPBalCZB0m%S z%U?JK;SY!DVNkw+zcQqR_>C9RieNc8MR~Bijwbl82y|s-b@cS4^<==ZU^yKf;4d!h zH_`v%LjEHcbmR=^XLd7zZ=-AW^bu_I5 z$5s-VaMtjbttgtYX#Q1L{{8#n_cdT=UD*it2N*?>!V0{OeF^8!qsLC_Ts}!#%_@BC zOApD@9(XNr{GskOM+8jY|M(SgL%_2K=L|$01PO81{Y7jfeEpd--z9)JsU{|R(D z$NwGgKxg1SLYph2&l@nK--uDKj6SxCiWqgJy2CEOD?Q#MVIX@(%j6~xk55Hw!ir$R})gK@ih z$3)NCQ_P~M_`3ZbW#xPK?#VbnLsi6lTD^Ho2WBFxEiJV$#mO*?J&8-x2_Ycgi_-i0 z=%TD!aIZ~bHz^L}$m!%@`>B$c7XdSqH@}w*yL=s1ZDO_|seGpb=^Uw*u@EH7FoRR- zy}6HEjukLI@FWg-^Ben0h`cabB8-Q<>Y_#o7{{faVPY~o)Qup%dDB-OvOjN)s*Hi= zQw{x}-7AdO#bj*|(%O{FCiiCeVybTPa$Up)1^Msb8o2m6J1l*FEEPNnNamVocDp7L zyD$G1Fbod~iE73tt$_su1SU**Ak3vgU>^MCXjt33Kg~D|lgUI5^6D)+R2~TUwE@X$*2He9D{1hMHM(@b0B)#SWIfj> zc5?lqR?<+-Bo$qyfVog?|Hv}`443gj>z_8!{|v7YIHYX z-_EE@;(#I&$5X#<{;O@VCR%7pm5rBWZFDs!i{9*6GoOG-*k z{osP4f2<3A(W40oUR@0h*?HvJ$Y|g7QF?QS-u|ks3n4IR2gr{&lk%;8PgxpSd|1+H z6YTWC?ZQ_6+#o+tC0Wh6a#h3fDe>6lIxY}08TFz=Jygv;ksWG1bw-CT#|*!@@Orcq z@9!qO5)R8q88ld3M+_iAVxZi~I_cct8x9T*_)n3$jx|?gtw?UoM)t0n@7pAuYx-K@ZQ9qqVY$Z$HY^ti9hY?z;268(8i@>C9FAE^Dst z+u7YYTx9=!@nE6ISvL4)5qhU=Q8S$7QhK*zy$U-U8#Y`4>g-$>I>9%;vAelkTzA)N6rH;pIl|sJh&P_*IhYJk3(kzD^jx#b%84KOL!#HJ%@jftk&poID2bh% zEHIy}f8yBZA14}BfRxg zR31W(J8K&!_q1SdD9+R6$M($A-O{Y_-ES+SK_@-WB{zOj`SOl_k>^w4X|<3gYfk*0 zVE9bCQjlB?u}X)SIl*}k#{m^)RK@R<DU z_2NQ5|HYQOo1+Co-V_@g*ALrymzCFk-As}?sy3~_z!7*oWcz8al zo@s__5|}-+mSf6pPZas+Q#l4>RoPCD8F=lZvRUHZY9_gjc5O+*QWIV9GaJf@Po7>> zera3maA$q^dFmhA^Q>x&iTfR5-D;6m0jHyf(yj9y^js6-58u7hm{x<}i!>xTOs+T! z3kWnfPG1tesCo!dRZYQQFq`YE6nqmutLh+X|KVBkkIF11&;>I^poyST@GPV1*Ae$D zoySw42Ny&@;qsVo_mVO*+nO0B7+7q7Oc*wV1i!FjCqjEVX^;*vg~_4^#)|}TraY0t zn0DA9hH&BpgN1Hv0=532<;E3{r>(}c(YI_Hc7AYh_ET=%G}Kt1cT>z)&rT{RM&1)? zvf8$qU@+WK%_%Hz=X8U9>5aj4#+dKX(a~`wS(cFQjk(Ua(`i9V`Dgm;vtLE2tN?vCn?1I0$RbxU@^VJ0@%1<#e@%6VGGUvpp4re)(eQ?%(UjRdtE@xwa?iz&*%XDw zmFWsSi2jf*b&wp->B8g2z4BoW6Iq{h$Et@&ByKa!@Ig?cWs11|Mpi56o&5$BkTU(A zIqZzwA5GmWuXPS7vst2#^*qk7e5$&?h5CTDhsC+zH-@e$diGcb8su3t%+hw&SLS$C zU=gx4j<1Dl4ME;dTKt<%pvkz|FRp&no!+?}I@NZRTP4+&-%{f&e!QtkDo(G!FLJ$& zoT~Jsm5b9OTT`3CdA1>_!gfB=-~zcYNm*RH3sE|nUq>*)Kwomp^I&rWQfs0472oAJ zn=cRugve-4iLw0Iu`gN8(e7N5@o6IxBG-s@Ih-G|ICI3_(vzF9CtJ|&Z@wJ-8FF8> z%F(H|yr=?_zWz}_ewa_b93qYjJ$?RzmWY*)T$4y#HM~Lj>IK(`IOb@t=`1#{gX!km z+>Zrcg9S?CC3_D)haC$I+`VkZ=Qd_E1w)?F7v?m0)-Q}Pk?j9xh0>8-&%DGkN7D2ku|u+ z4en;~tfbC8R+%FCajEIMbaCrspx+9BVoRDNOuVYq)X|)}n`pU{Bsxx$43`6>m~oLm zRU{Y~Rx<*VI&XjQH79t;W%~jx^Zgu_Mhm!c%4qpDv$=|;$|;M#U4qno%tDHdodc-!;rTM5U{$6q$`18=H`KVRWYVB)z^EgTeD zn_C}}p^wgl}Wn7%G zKVYTZP6a6Q4F!|X9f|~#Fmp>bA{|Dm?AIX{hyW@RvEzPYW1s$D+o@ZiG2Xyok7fQiCVGJz1F{;x&=X0wn zYO*-ohq-|n?4Ch zLgzS5KFiR3*s{qc+7r3Fw{dTHr{=>^Ng$pfsYQd{3yb-^v)?lESR#q-5p|}$JWA)L zb%LlF4^_&}TKOv`v%GqppOb!rpLI-QZjM84bro;fGOh~o9O=s-PNT(|_*cas^m)>A z+acsF5e#%NNB2Ckk2bGp&~UIHmYM`E-J=CeuoTQkK8}jY{1$~74ORI8YiPQEVrgi% z;(26bBzrom7h|2+nbv0Wo7;0lAM~@H)Zqb=gOmS#3xqSKpYr7jG}?*|IWThYvR@?D4`ucA-IEbw#q1XyqVuxQcMvJ>|-?^@nr_@}M`WD$XS1T=lIx(Vh72 z1m)axI}9tiH7)qUbLf<*x$5bY2+l{w2kwp`tO zz3U}+!?Z+Pr?!Cr>dlFM=$jmbc3yTLa`Nq5Cm%FnPQa4LM)ELja4OzldzR{QW8;ToZ}j<>kBz9?j&gsrt5 z@Rs4R9yQ5=;g&6WrR_Y$B~E-4b%td#S?#vWl!5K1J_;#dVDHEG?}xOK)~151g*lEg zB+6=l=4obTf7(QKC!jX|%OAE$u|b%)#f?|?+ zFR9l{ii@w9sW-3o@9hPPN{VMr$GFv8NB^n%Ll<)^$il`Z_SKLV@tjy~#)|+^b|FQa zl2+hn%Jg);&u|et??ugg`cq?BCOKuc-j9tSua>mKSz=$rzFc~8NzusjJ7#gO8cWF{ z`Gdb5hY)XxM;-l)aZ1);{eoTkE4J_J>R24tPc18xJC&>UA^1`H=VerVzDn=@uwilA zNH(RCiA``s&VlyD@x0l{496^$YD=A;GbAtHJQE`wKm?vQ>!A7H&NaCs_~a}t(KL>J zUUZAM3vA4D(QWHB-WouO^NylA#yhl286AE6d}HNR@q*E#)FzY#nz6(2(zx9(Q`|+4 zczGxuP8ln2o-R)@d@iJ{;LAfT?w+2WzRV+!lnW{Ix3IGV(DxRcf9*qV_c-T>pXY zef%lb%RvG{%H$8=l(~{tzRGd@T3tEeEQGC(x>qh3lXe_<_1lRm;#hWuFxwjMld1=_ z=t?(`oG)(#9$gkO=T2eV6K_)^)a|GC2R!ChUY!`r?mXZeNr1dipF8%ppCy*Lq~+fB zVfud2IMObm7yC!E`9k&SD)4UDw4&Lal)0Ml$jhT;$3J^U8<-n++li=y8U~JOMj}qA zdjhMVDQfNv0)kibxaASbypKXUz)W=Nk!5^VbpJj*`6c<^&4P|*ME|>4(9w+Ox3i!S e*S%sodb+1#Q>rIbPtN}wG8^ie=@e<*xc7ezU4{Ap diff --git a/content/ko/case-studies/haufegroup/haufegroup_logo.png b/content/ko/case-studies/haufegroup/haufegroup_logo.png deleted file mode 100644 index 5d8245b0f6d18fcf18d871f7c657716b37f2ef96..0000000000000000000000000000000000000000 GIT binary patch literal 0 HcmV?d00001 literal 5794 zcmcgwcT`hZx5ok~MMOaq5FtuaAhZxd6-f{Tf;2@y5J+eN0wIJBib~4>(pv=S0s_*z z8L0w>-s{j62puB*#TjS5_s93Xf8KiUu6^%0cdz~1XSaLqSv%w*{5~@SCj$))4YP(i zRF66iQE&fqbkwJhLD!8yLD|FE!U~1N@>(Ko?3@$C|e_a6TY8ZsIoE}u|*ILv!MSv|9>nsNZy1ToJxl4(mF*ZO62m}HYmjp^mic%#+ zT|J$!79OHbt~dYI07bf5VeFi-c4#NwpBgPJ(Qa5p0M*jJTyS*O*8WGalk2ZQQ9}mw zuy6)Sh=~Io9e?`ur?e|p5BV=M{;jmDfu}PPsE2e#yJ4)T`LMqEH<%i`|GdyoL8>)! zP_&hsBhm@00aXN0IbsMq1T{LUFj=SyR1K^OmXLsfA(HoGRm7$5se&OgU}-6-zp)xl zu2>5vE9Bo;yMJL-|07mT6@#?EqA><&w8P)oduWTsqFrs#&b+FsKeGkm71Fk_vUB>$ z6#g0DU-lv~b~q$L4TE;%{WC&xcK^l-2?$798UhA^p(>z1J%C9`LDkeG)FeStAZaKR z@C%FhH{Sok%KVR5Ak`S)&-DJcbpOqw*3i%CKgCOZ`KSDmPSl#mP>WkX;0Q=VbNQkM zRK>t!bS3G$hk;>Ss}5Q7cJ3`i?yFZH&H_-wuv^opv+Q6tZe^41L{?VkQCjGAnD<^; zdGYn%Pr2D9mBW7jC}e*pY~>4__Z9#7vn3Z#YasLcKYgtR;$i3bLhL@)o+uO3zvfd` zYdiK*T~|GI7Mx{$&eD9t&Ly0t;T8KyG)=U>=y?^*Uqo|};}_A)oc=|3O=KJ}QSib{xyoqC-oEdTyOhm^qT6c9F?Pr8 z?02H1q@>#34Mz1X^3|K^6xHpHR`mDx>ymVood5WJk22{Co_yC6*)>F`!?zXLy2zjP zNK?0`r@!{`npK3BN2Xdv&$uV)DvrEAJqx=rv5JFe2zyVR5pKikMFv|lvKpTUyJl*7YT6@gu4kZ+_aAG-;EIx&k*+ zyDXv1EhK=xaO{!H$2UXPetuCf|Kt-0KR^Gcux0SM>NiO+z1A5{$!E{;%>(w*6p1Ov zk$cpH@YmmiJ>n}Ao8BHQ@}KZoDPMUN4kyS;WEAElahtK&%`;#&DsyfIqFmn_nQHr+ zh{WBD7SyhBJ)n$zX>-i;_{tC$JF)ov4d2JU^^z(%2Xzk32RI#{?VcB}IT*a@=(!4{ zY)I{45%`-yeJb?3H!jZfaxu|?07J~(_T^^!F(KSCT-O2uzWJDmy{4nDoYylKPZj(C zmU$sKx%vhWuQstsm{PKR}+qJe+s*~eJz^}<*bmoXdHjhSIuLM2ieu3VT=o-N>b_V`b zS1Z0?iP7Z`Sc)RWbXxf)S46y7ozcO>)UjII#`)g*`jv5RTy3d$1e@lwXBY3Y+9VI^ zeGVwfEnsh0`dE^emxp8Fl0>|!P7lDXj+EB>8yjDu}F2uB%zUWxuHd_$c2jaPiQOsT~BT4&?- zHrY!W0sA|s zDntA2-E@e%U%M2J$`rvs^IET+oP+dQ9BI9@y5W0YHvRQ+VycYe@#Zrvt*_T-Jlx!N zK2`fCj8Vplh=YOH^V%%4JGlkr-SZ`qobD&SAYlD=uWopS9r&r~@wq;x>EI?tHt(1E zRvGE(ZDYIH$<%*$Be@V**mO2)Y~JZo-PLy_OIEWf=e~THU(eU+KZFW&^|c&+i(3?) z6O%Sny{nq;r^Rj%!}9GCMUB3sGX2?<7? zKKBmQiMNaDCEP9ncGs^GN9#t|czB8@Iu1xO9H7iyMQQAC_X~&d7v$?4YftO#!{YQV zUCM(*F|AJpJ}DS9?E&*~ZiwpX96O1KwE?svm4ZlO-hX_ik0o3%W9JL!QS`Dr4q}v? zOxnucD+ikqGd6xC6kFL1v>V>{gM1G;b+YxDUZvFBGpj9r<4R)#04!vfx1eh-u%Ik1 z{-!EIn+N!&QabyMSWHrqy@f7I?Agyl**Q7-aCqmcS-n@7$6Q*_exV7`kIP(sX#tPd z&o~zya^bp?mX5;iuuyT3z3{+gVZ|T|OmSL5diobqgDtI~cdTNx-`)?Q)ZzCB9#hjF z_02w4%j^i3dxzetYuVNjh@QZ>(4M7x2#Zae^f?Oq+;HBbUddkbyuk`E|4zQJZ@viLjVWyt)?lr%9*`lY2ygY$WLoA((xvO`7^57%_&*M|%T z%)&9C{SyQ4c#$W)VK+zNPAG&v8?lqH&)G}KS2eEpt(tJ^aWD0^ClCnS$`*%>3nVV^ zYcbPxi>3Uc6N;!hqi57BK6*~3^ymwZ{5l$EGj4~eSu3Zm){(fC1%AGuskz=#|hOh9OQ*k=4idS)tz3qK1?n><-mg{;pg$JgEEZxzEf(Ar6 zuE;$5;YLA$z-o|jfk~#zr-~JQPag{l8BS?LH|=qL>&|X50Mm?HUO3Ol=sH5aV2?t? z5K^*Q8Uf#&`MVaQcMB&2Enh$p2*gaoYesh=&$WY|bXYOtLYL-B96_kpRljsvS1Oug zB>1Rdk~pRMd=)KJUeIDAp18HBgNg(z9L%TfB(P1#!;wcaPUi zfnlZ%&w*F2^fcWU?5>01H>EkXhj$d{@$0>HEK1)_4l1Xnr3JIg4fGdWFi?aA?X@sK zkbpk)Y=QB?$mHDWguXEzTDenGV7n6<`x5K*bQwYvKPT+D?_&0(X@=#giQe=sfPwS; z#Ek8C^GB->v>rSl-&M)bL>;kSJL|ots-vvPUszpS;o@(*^ksi}+bi)mbgOV}5-erF zh=1-_?G`_ONGMm9P3GDOb}zK)9j}#1>uq<>@yvmc#bhP4=IlVR)y^=tXn0xaCL62U z>skA8ukomWz(#V6z`dNl=88w3*o4983vH{oiI0aOy=g7m)U$OtTvO3&7eP-Bb08Z} zeoX26z73WU=zQB$Gr)1(r&l05ug3UQjoe9|CEho>@}&8S#Fd~%wMc>Jx{Sc_B<78l zl!lvt6$O##57A%o%$25WXZ-5Yv=}mvanW&cYWsIz3d%1hOe5go<vXFfD$|J7Ogpm9l;wq?PG zOzDAw*&vDms!;s?xNSlhySQhgl!%1I%*lgLd=?mJobh>VqT9W-xm7s7y+gx>DJrED zc0cJ`=>RH#j)^Bqai)`1&~weB)eV{s+Eg2!MVUuJgY~xY_3w%gwvxtOWQ7;%if1x<#T92j|&|AzQ{vv_(6hAkVjOr}jgI zr?~^&efp_w^QyYkY7J{yL#gN0k)!nXB%Qzk;sj>F@lZ=>>rg@(UGBj0UqGH|+6 z6mzXsd6K0ZzCF})G3JE|uHFf~)&GXmm$-9a0O3-dwPdJVwws1Ln({jeNJ$2j9+KTB zFO+TuKw&UlPgwPea*9t0+VbA%ofWsSkftUmLpObY4qctBwJR^Q=sRiqsxm2a;he59*e=O^3V~JtI@;Jq1D5iQ0Yj4c8Wr_9RZ zMhUcx9JdYq*yZcZa*A?uSz#ic_-V>$AZoKFH|d>td34lqSB}eHiLnUwGcP zFelw6R&SxJuYU)=2`dj@!5x4c0|NuI z(c@Jvtw)rpRHc2645^j~f7{dn`!Euz>&7Rq9sB(Ww4XNy;pLTXZv8BQG9Ba{ktzrA zS@@c$TNkJssTG+3gI=kqK^?Jrx{PV_VQ(Y4xFGa zONa$whYFucqbKVQ{k>NtHkNdQdhaDRPTNu}IoQ}%#>=g8R6Mk&(;RSIJKuU#6<%I6 z3=JOS!LJ-l2bK69P@-Qj56o_C&a{93yfM4n(9p2zkao`bma=kcC?o!TRc3-18)5N_ z%ZqVBi-DnGmK}&wnEm49^(97I2Z}}7~CT3=GQA34nJl9n0&d|xGYh}L}8qqW? zk$7}N!Q*4*hcW~AAM31g(eav#ohjO(tD^&77|m0iN)p84VvH_DM`7H2%zG++yn;a= zp1xo6;tahU1je;sqd;r!zok77StY`~*gN?dpi; zas9}u0kG`~*OUcEw33r0cM@e~y7NPI54s+$Lf)X_7!=Yu6WQ3+DOri>dsniT5z_qB zmfIfv1#a;Y@uYCd1MQtjkTuhUl2272Wxk%Xv%U^TO0wMW8qREseoHXBfXXG!%*>Fv zWE{l(%A9pG#qOd>S0N?1r82wln3x!){)RgbWs=E1l0c~O<)NbZ#rgTy=f=KGLdf{arY`Um!zM*px#rDcr?bbxUzAen$tgYV zVVv@-lcW5aC)~p$J&02RORmqBuU*J}7uXPUb`94eS;^{wZ sh;ES?(ecQh|0z0}WWqG#+A{O}&#tis3=S>4_blMQ0Oo0{wEzGB diff --git a/content/ko/case-studies/haufegroup/index.html b/content/ko/case-studies/haufegroup/index.html deleted file mode 100644 index 580a282683..0000000000 --- a/content/ko/case-studies/haufegroup/index.html +++ /dev/null @@ -1,85 +0,0 @@ ---- -title: Haufe Group Case Study -case_study_styles: true -cid: caseStudies - -new_case_study_styles: true -heading_background: /images/case-studies/haufegroup/banner1.jpg -heading_title_logo: /images/haufegroup_logo.png -subheading: > - Paving the Way for Cloud Native for Midsize Companies -case_study_details: - - Company: Haufe Group - - Location: Freiburg, Germany - - Industry: Media and Software ---- - -

Challenge

- -

Founded in 1930 as a traditional publisher, Haufe Group has grown into a media and software company with 95 percent of its sales from digital products. Over the years, the company has gone from having "hardware in the basement" to outsourcing its infrastructure operations and IT. More recently, the development of new products, from Internet portals for tax experts to personnel training software, has created demands for increased speed, reliability and scalability. "We need to be able to move faster," says Solution Architect Martin Danielsson. "Adapting workloads is something that we really want to be able to do."

- -

Solution

-

Haufe Group began its cloud-native journey when Microsoft Azure became available in Europe; the company needed cloud deployments for its desktop apps with bandwidth-heavy download services. "After that, it has been different projects trying out different things," says Danielsson. Two years ago, Holger Reinhardt joined Haufe Group as CTO and rapidly re-oriented the traditional host provider-based approach toward a cloud and API-first strategy.

- -

A core part of this strategy was a strong mandate to embrace infrastructure-as-code across the entire software deployment lifecycle via Docker. The company is now getting ready to go live with two services in production using Kubernetes orchestration on Microsoft Azure and Amazon Web Services. The team is also working on breaking up one of their core Java Enterprise desktop products into microservices to allow for better evolvability and dynamic scaling in the cloud.

- -

Impact

-

With the ability to adapt workloads, Danielsson says, teams "will be able to scale down to around half the capacity at night, saving 30 percent of the hardware cost." Plus, shorter release times have had a major impact. "Before, we had to announce at least a week in advance when we wanted to do a release because there was a huge checklist of things that you had to do," he says. "By going cloud native, we have the infrastructure in place to be able to automate all of these things. Now we can get a new release done in half an hour instead of days."

- -{{< case-studies/quote author="Martin Danielsson, Solution Architect, Haufe Group" >}} -"Over the next couple of years, people won't even think that much about it when they want to run containers. Kubernetes is going to be the go-to solution." -{{< /case-studies/quote >}} - -{{< case-studies/lead >}} -More than 80 years ago, Haufe Group was founded as a traditional publishing company, printing books and commentary on paper. -{{< /case-studies/lead >}} - -

By the 1990s, though, the company's leaders recognized that the future was digital, and to their credit, were able to transform Haufe Group into a media and software business that now gets 95 percent of its sales from digital products. "Among the German companies doing this, we were one of the early adopters," says Martin Danielsson, Solution Architect for Haufe Group.

- -

And now they're leading the way for midsize companies embracing cloud-native technology like Kubernetes. "The really big companies like Ticketmaster and Google get it right, and the startups get it right because they're faster," says Danielsson. "We're in this big lump of companies in the middle with a lot of legacy, a lot of structure, a lot of culture that does not easily fit the cloud technologies. We're just 1,500 people, but we have hundreds of customer-facing applications. So we're doing things that will be relevant for many companies of our size or even smaller."

- -

Many of those legacy challenges stemmed from simply following the technology trends of the times. "We used to do full DevOps," he says. In the 1990s and 2000s, "that meant that you had your hardware in the basement. And then 10 years ago, the hype of the moment was to outsource application operations, outsource everything, and strip down your IT department to take away the distraction of all these hardware things. That's not our area of expertise. We didn't want to be an infrastructure provider. And now comes the backlash of that."

- -

Haufe Group began feeling the pain as they were developing more new products, from Internet portals for tax experts to personnel training software, that have created demands for increased speed, reliability and scalability. "Right now, we have this break in workflows, where we go from writing concepts to developing, handing it over to production and then handing that over to your host provider," he says. "And then when things go bad we have no clue what went wrong. We definitely want to take back control, and we want to move a lot faster. Adapting workloads is something that we really want to be able to do."

- -

Those needs led them to explore cloud-native technology. Their first foray into the cloud was doing deployments in Microsoft Azure, once it became available in Europe, for desktop products that had built-in download services. Hosting expenses for such bandwidth-heavy services were too high, so the company turned to the cloud. "After that, it has been different projects trying out different things," says Danielsson.

- -{{< case-studies/quote image="/images/case-studies/haufegroup/banner3.jpg" >}} -"We have been doing containers for the last two years, and we really got the hang of how they work," says Danielsson. "But it was always for development and test, never in production, because we didn't fully understand how that would work. And to me, Kubernetes was definitely the technology that solved that." -{{< /case-studies/quote >}} - -

Two years ago, Holger Reinhardt joined Haufe Group as CTO and rapidly re-oriented the traditional host provider-based approach toward a cloud and API-first strategy. A core part of this strategy was a strong mandate to embrace infrastructure-as-code across the entire software deployment lifecycle via Docker. Some experiments went further than others; German regulations about sensitive data proved to be a road block in moving some workloads to Azure and Amazon Web Services. "Due to our history, Germany is really strict with things like personally identifiable data," Danielsson says.

- -

These experiments took on new life with the arrival of the Azure Sovereign Cloud for Germany (an Azure clone run by the German T-Systems provider). With the availability of Azure.de—which conforms to Germany's privacy regulations—teams started to seriously consider deploying production loads in Docker into the cloud. "We have been doing containers for the last two years, and we really got the hang of how they work," says Danielsson. "But it was always for development and test, never in production, because we didn't fully understand how that would work. And to me, Kubernetes was definitely the technology that solved that."

- -

In parallel, Danielsson had built an API management system with the aim of supporting CI/CD scenarios, aspects of which were missing in off-the-shelf API management products. With a foundation based on Mashape's Kong gateway, it is open-sourced as wicked.haufe.io. He put wicked.haufe.io to use with his product team.

Otherwise, Danielsson says his philosophy was "don't try to reinvent the wheel all the time. Go for what's there and 99 percent of the time it will be enough. And if you think you really need something custom or additional, think perhaps once or twice again. One of the things that I find so amazing with this cloud-native framework is that everything ties in."

- -

Currently, Haufe Group is working on two projects using Kubernetes in production. One is a new mobile application for researching legislation and tax laws. "We needed a way to take out functionality from a legacy core and put an application on top of that with an API gateway—a lot of moving parts that screams containers," says Danielsson. So the team moved the build pipeline away from "deploying to some old, huge machine that you could deploy anything to" and onto a Kubernetes cluster where there would be automatic CI/CD "with feature branches and all these things that were a bit tedious in the past."

- -{{< case-studies/quote image="/images/case-studies/haufegroup/banner4.jpg" >}} -"Before, we had to announce at least a week in advance when we wanted to do a release because there was a huge checklist of things that you had to do," says Danielsson. "By going cloud native, we have the infrastructure in place to be able to automate all of these things. Now we can get a new release done in half an hour instead of days." -{{< /case-studies/quote >}} - -

It was a proof of concept effort, and the proof was in the pudding. "Everyone was really impressed at what we accomplished in a week," says Danielsson. "We did these kinds of integrations just to make sure that we got a handle on how Kubernetes works. If you can create optimism and buzz around something, it's half won. And if the developers and project managers know this is working, you're more or less done." Adds Reinhardt: "You need to create some very visible, quick wins in order to overcome the status quo."

- -

The impact on the speed of deployment was clear: "Before, we had to announce at least a week in advance when we wanted to do a release because there was a huge checklist of things that you had to do," says Danielsson. "By going cloud native, we have the infrastructure in place to be able to automate all of these things. Now we can get a new release done in half an hour instead of days."

- -

The potential impact on cost was another bonus. "Hosting applications is quite expensive, so moving to the cloud is something that we really want to be able to do," says Danielsson. With the ability to adapt workloads, teams "will be able to scale down to around half the capacity at night, saving 30 percent of the hardware cost."

- -

Just as importantly, Danielsson says, there's added flexibility: "When we try to move or rework applications that are really crucial, it's often tricky to validate whether the path we want to take is going to work out well. In order to validate that, we would need to reproduce the environment and really do testing, and that's prohibitively expensive and simply not doable with traditional host providers. Cloud native gives us the ability to do risky changes and validate them in a cost-effective way."

- -

As word of the two successful test projects spread throughout the company, interest in Kubernetes has grown. "We want to be able to support our developers in running Kubernetes clusters but we're not there yet, so we allow them to do it as long as they're aware that they are on their own," says Danielsson. "So that's why we are also looking at things like [the managed Kubernetes platform] CoreOS Tectonic, Azure Container Service, ECS, etc. These kinds of services will be a lot more relevant to midsize companies that want to leverage cloud native but don't have the IT departments or the structure around that."

- -

In the next year and a half, Danielsson says the company will be working on moving one of their legacy desktop products, a web app for researching legislation and tax laws originally built in Java Enterprise, onto cloud-native technology. "We're doing a microservice split out right now so that we can independently deploy the different parts," he says. The main website, which provides free content for customers, is also moving to cloud native.

- -{{< case-studies/quote >}} -"the execution of a strategy requires alignment of culture, structure and technology. Only if those three dimensions are aligned can you successfully execute a transformation into microservices and cloud-native architectures. And it is only then that the Cloud will pay the dividends in much faster speeds in product innovation and much lower operational costs." -{{< /case-studies/quote >}} - -

But with these goals, Danielsson believes there are bigger cultural challenges that need to be constantly addressed. The move to new technology, not to mention a shift toward DevOps, means a lot of change for employees. "The roles were rather fixed in the past," he says. "You had developers, you had project leads, you had testers. And now you get into these really, really important things like test automation. Testers aren't actually doing click testing anymore, and they have to write automated testing. And if you really want to go full-blown CI/CD, all these little pieces have to work together so that you get the confidence to do a check in, and know this check in is going to land in production, because if I messed up, some test is going to break. This is a really powerful thing because whatever you do, whenever you merge something into the trunk or to the master, this is going live. And that's where you either get the people or they run away screaming." Danielsson understands that it may take some people much longer to get used to the new ways.

- -

"Culture is nothing that you can force on people," he says. "You have to live it for yourself. You have to evangelize. You have to show the advantages time and time again: This is how you can do it, this is what you get from it." To that end, his team has scheduled daylong workshops for the staff, bringing in outside experts to talk about everything from API to Devops to cloud.

- -

For every person who runs away screaming, many others get drawn in. "Get that foot in the door and make them really interested in this stuff," says Danielsson. "Usually it catches on. We have people you never would have expected chanting, 'Docker Docker Docker' now. It's cool to see them realize that there is a world outside of their Python libraries. It's awesome to see them really work with Kubernetes."

- -

Ultimately, Reinhardt says, "the execution of a strategy requires alignment of culture, structure and technology. Only if those three dimensions are aligned can you successfully execute a transformation into microservices and cloud-native architectures. And it is only then that the Cloud will pay the dividends in much faster speeds in product innovation and much lower operational costs."

diff --git a/content/ko/case-studies/huawei/huawei_featured.png b/content/ko/case-studies/huawei/huawei_featured.png deleted file mode 100644 index 22071b4691e3830c0ee3636f04e6aabf02d5fe83..0000000000000000000000000000000000000000 GIT binary patch literal 0 HcmV?d00001 literal 14310 zcmaL8b9AN6vnU+fwr$(aL=)TDv29Ll+qP{@Y))*O6MuQ%bAIREKhFL3TKlP|dX>7X zc6V3p?(m=TlJGD%FhD>+@X}IZO5bqo`040E-iHE}&fDZ@=l+;2+!&yU4mdDu6hQaV38U}Y8`+wMh_ypbU4UMe; z&O}B4GYeaOlIzZ15+Vx|eiC(dIVL%KQGmIHl&2#=*;8J{*wf0G+k`|=fQZkX=Uadc zz}b+<-NxG1iN~FvJ(eF%fZc zGca>7v9UAL5wS2ab22iqGcs||Gcofpaq+M)6aDvx zc^Dbp+}s%4SQ+db%@~=vxw#pcSQuGY=)WoGojh!v4c+N&ok;&9K@8wz>}X-{Y++|h z^p8YCBRdyoev)rX|Emi&_WvW+*6F{8>3hH!-3{#-nHiY=>C%4)<>dbVNo{QYhuX(9Kco25+|Jp~$=uGK zNK}-FO3u*O!uB7W`dO?>(4jL8DDv+e># z73~I0oKK7k84+EW3kj6SF(M6p@m`F7#vfT&C@=!5%smhoOc4ou5}PS60gP;rf@50# zLglNnBBrqsDr8&t)U5HzIU}d0cKc1Qj3!Bzf?I*_1^?GGK>#H|3*T+sVfcSwD80hJ z_&?CvV=w%_q2UDTzxX#q)Xs6=`!D!jF7`kGj2z{_SCgD-RXlvROc?wgILqyWa1OE- zLqBGdS}j@()s|=9?D~tCs2H4soQIB+qc?UkMa6dy0TmX%PA1(C#>_;0H=jx z*^1&p#v$yZ7)kr56hlm&=h`ph%wS)7Gof#}RJ=c)e zDi27LDj7-+Z>Q-Y$`I_L`z@48WFM*lFo?UxF7tAxoF4(e6hDxAaloncbPftA$q2T_ zA4wc5l)K6zt0Y3Q74eBHb;jbt;YF~qtkVr$my@sSot^BJYfN&Rv18WH86TOpFJxR$Lj0C52LBG0o;&tAaK7p z(I{(W_`yZE0xDzekDe1q+A!TWZt2fDjBu0yQdM^`@CY!IBavlbL;uE`d5`fU7lM}0 zD6sR=>SAL%agt&+h`6kptY9{pz%qQQ`5!XAzk8!blCrhq$=dygVJ8RX#fu1qL&Lp^ zkbL~gJA4{Ls|7JfVAL^V+XJYuhKwO-Cfnf(EHJA1>r!-{A7`+J_Ixy?Q4D}b3W}oZ zIcJr`EaH3l;&_IdaAtyhqPiM?)^hC_o+}hKB+YkzMjwLB5n|M{B&8(UV5f=5dWH9_ zo+<2X1bKuB)=*-Bz$jn>A+r#X6NR79XpD$?Ax|O$e#3UnV0wIjGo;_tJ(|UxRJ0~) z_edtS8(UY9r*^{OBBz2d>&+sdoi8OrLR(Suwjx!A!f`1H`Ug(#Jfd0dxS#+`8+e2l zQS1L6Uwf=t{ta}P(w6xvG%H=nW%^g?1Y>XGCXikC&x>-lYfA+Ph}bSU z%#Z*}ANH3bKK0ANWhR&EZQW9xI4yO&+dI@6IJ1x;E*Dm3zsUf+cCNedR#U!wbF*70 zY)+aUK5tD3_qhz$vA{3;A_>}!bHhqQnTy6)Yu6a1uw-x&NP)WI-oB*pF??+Y{3Myp z&PW}QM5?X;S}sgA=cus)L)F0!1j4a_ypVdsWN_l!O&mmR_%*@BeSPl;FzEdx15l1H zELy02=U7PK?SXo89usQITfGcgwk5n_cyLrYGz}9&LYIbY;7Ad;(X56y%B(Q2Rm-~B zit_k=$9R*NeD({Vpv(p1V;WItW$}&*^h=q+JtO8et`7uHK4p10(To_CyHrwW!ssd{ zgZOT-V&q{6^*8{(&Y@xDmPMf`zmn>sRI}knU2%i#iG5Zymo+FH?`?^3a}gJOG%YP~ zrCAvX1ZGJ+QvjwkX@&FLXh-=ZluEU7*j!cl=cQjmXth3Q)t!U!{&fu(8Hv88JU{z% zL@LRFGh{ZEQBqhrt+h7K5UO4>gm!TW0gGHtc~hA1FLC2Y7{YXrLfrcD-_G)ieJlj? zwX2<-7Xh6(I#N};hLx5_sPVDtN%C!!FOO@Yb>YZ^m&da1={cF@rIY@t!@uiakWuA= zkP_Uk$#$e1P@b=ekC~;A?za|_~x|zX;cTJ+|xUzemXA0%KgO~P5pxoe@9x5s`Y~316sxpcn}VZoiG;*; zM=tbdI?=-2TsPUHn?{75(PxW{Q(R9H;gl7GC;`$I=(R$?6jS{q5QC4Wl_H-KHsWg% zbq+#l3+%dg4bM3-ZYCnGv4Xt?)Nn;H#%eD%6(1j(g0hP(*+^pJNp}piBv7TZu$@@c z>!|%{P4VFA0s3nFb5OfSI2@yi`Sr5famXIYLJ_ROzhx=+wGRw5kuzI)MyDv%{Rm#% z(U0$0!v;TY0)8G|oSwy8e)mkKJdGvC{erE_mf0Q+N|jMN?(an+kDs1=Ox!|RQ#U@H zcndKbH7`eD#;TGrU4uWhwmRB{8dFZ-d%DEWC6}NHt7RDWNE1WXL=}m@ zpe{IrpvCAjvb(!Lo6A*r^mMd7I|y_W-}-RldtYeo=6Jgb$?0$*DHB^6N=vlXX>5_4;lI z3{}-p+bEC#9*gCJP^p}CmZ=VpT&Y=_iGi7!yShhh4N*-7d^gTC(CKU)4to*Se!z1~ z`QfJ7Uc`2%1@URX&iq$ zqsfALD>3@#v()CpgqgT1Dx5d;=V?y~m&LdJGM`^A*^ps7*4jbf1lzsG|31_vhE|KB z*=%oBs2|1MxlXrejRqsWvVveIGfvIBmO?v5RL7#gYWY_%k(4#j&F!jw|#QA}9%D_%3+N+9|O9 z)gnS_O*>)28WphT6xjulS#`!Ib)5x)@7poMxKGn$8YPuwnq>_SDe_1qR7So|0;G)? z4fcEgfGbMuKcu7d?ZI%aTG@`9s~ zHy0E+N@T?S>Qy$ASKrhmkPLu#fzWmE*bK{VP$PRYW~TLq(UPpBD)d359x1>k7?ft) z)9A2(cR6P~hRYp~SiLJ-%YIF)6pZzG#C}I&g$SWv0#qtZBp!k{o7dj$Qy^;hilOM& z&Lv7KNvEPHxvku%6A_cO9#se5b=WD@pz;)Ama#)tWf3(lC0Ezi*(!Uj{nn7*v0vrV zMzUuKFXw0o3A2~sgu-ieLJ-v2!Jd0FORb@8gxrSY3gFRT;+`yV^fD5Ke{&=?X@LI zi7f|%s-uR_#SPM$#FT0*R=#*}F2hZ0Hr1MM8?I?CKJzeXEgExS2XY z80OtTQopveFEHDTOE^BTn8ft~#CvLdRGL0$cMgZ~^vhU-oT!Qn6)%6igLKqZ4P=cx zpoOB5B?<#3h=1C#TO0scq34TXPM!)vSCy$VtD?@+;>^L4T&AXSV0g~-sHRH3jK(nh zix{C;SzTTP!t3SdNnhKSmN^kFD+>-hmPiLxM^J6c6N^$AVq7B)fmV_lqGSp93{NQ(xQg*uBJF0eYkuo;k8{rq;mVJaH-S)yh z1B~SEPJU~AiWZfnJ&oynp|&J~IfUN1_~;{URqF0@{pk~<&$Q>0M~svX%rX0IBgqL?o){G zb%~#bXsXu3B+5wBq`ES@o78z8Q^U=HH${upPo`EImCgekK5Wx7+!SVdQ~{ql+u~z8 z4p`(d=1%w6C4c%LM>&iYgu9r%fu?(&Wtv)5)UU-AM`DnXzgAZcem5m(K~NYnor)Im zUjwUbNg1f258((L6ua$ftH;>;o0V1Niz$rD+up0Fw^NrOtE~}FCNc0(6hN=5SQivM zN5jym@khnfb_pCpC8b!+rpVRxB>$+m9p^kU{|_{f)>xUL*=Uk#Sw{sH{+xhnd<@Up zci?Vm>M2XaBJItykX`)bP^)1Az*7Dgs}?Ndy)^2$*d^8Nr*Ew6_9{%5EfRyoUP7vH z=qll@NVKR!%WOvMxI%C{g{TcB-LIEO)dd8r*cX+U4&$1H!BEyYlgdp;cspB{F*5{X z6G>VtAd!d0Ki0XJyQQRKpR5M|YQYasruC0d$_u$1xt!!q=q{+en0upBVmvrDdpavs zUOGH^)v7#6XVn$IFj%HIbVDZ97ROeEhGC!5IFXc1PN_AfJYP|r$+mH^U^9rA5{Qye zkif5JVL6~lb)o3m!K#-Q(BJtpG>HDI0YVpys0^J=YQYRWU&Y7tOGGir?F&uwwWGOs zZi#cX3$D1n9l=teDRo+FO{^53eM$EslT|YnUy?mBY6tQ_k9S-t_N;|I-Ys>G#$Dpb zxiy~MM|9*T=;+JBq!651Li#ZR8R^7#!l`qcy((?Kn5G4j+THl7W)PY(5aWiE9@L6G z3j2%3RBow~Ms7(pZs{3hM?s#8$LYtko}jDySh`jVQ2zH0j$|L9Yz`c>`@UH~o~I|s zp{9bs{Q|fyWcB1s6hXqdA8lCHrTuU2y-j~g1qEjc_szhomB$%t%TJ=&gO<0tEPDi) zy-^iXmB>(RQRMc}GDG4BP~$IS&EHwLw!n-iiSOYS64mk_qb&lX(42Fl)(OC^OaUOO zOgLs{a8qT!Z2yLYY$Y82ZrvPC_mzI!(6k)$y~FQF?{lSV(wUqaUawZl4w^j;*Vw1E zJHq9Bs(XYbclVt6*{au}s+4657n-z!(-}g}qMC6cWBpFiU>%NrKHmL=ktJ&|&F_rj zcuEhOCl;@v>7>|aD!+DeY%CWb(cc+~_bJ57; z>YUufrj|hvl~&?eWLnJLk@RmSRr_ic-}3d&){-r}Mi2 z-Xb*kt`2Zsw6VgL? z#|Zw!(^&VNj%#A12-MPlL>X349aJXno9?#};u09vezO`c!;@M^K?YQxDU;F|rYw^g z{qbSkadP_O5r1emF7Cy&XMr>k6XrNw#_0M%GL4U9+6t2+{Mj7~Es3u@gL@S_g3%p0fC<;Tk$Rxt1CJ>9$vFZ*DaB{j=@ zm$zC=zE1=!G8X$_uKgmmF|DF;9KEh;14G+to+LUtIxC@M4-jpFa14;tWm?1wv?ki@ zB(?!PfsOZvO@D8$PuxhcdT5~?<3iyaDZaKN*ZYfZt2a~8_LewobH~Jnh}rYBpFNCE z2Tluc+!_g%lLb@{t{wpOH;d8?r)84Y(dniq#)PqKpn6AR3e7~d+f6B>tzj)l1#RFnDDAW~;uahm4!a zHL|SdCCorEG7g!)QC9({{qYGJ`=)T7^Mp;K?Mv5bCVL9nniwGcE?l67%rgqLAPLhl z-tt*SOi;#1T^|&R0;OiR{xBGwdYS#VtjmhBW~p1*+Q~Zd7!#1?%TW&?`<-@1QQ#ag zG-kFlXUwg~Zw9e`em&3LIRnl{U?};*O-nlV0QonB!UuSN6!Lv~0BpT%?gU2-@Z~^h zqQ{qH6H~2~f3Ye5TSv~UDt_I<{)vsbE+^`3J$Kdnb)#?FbI~)puEWgjZ@Z>=YgE+N z4f^$S;}~`sD=l^KTa0@TdMs^5MckGbkG;$MZUiRVn0}YhSgbnYrr$HxOKCNADU{rO z=Ej@-Lg()Q-jYfdNALorWGvUE8LT*>&8<{=rg1x@@%BSkG^k>$dUA7vJR*Z4XGGgIQdi3P+_9jNO$kC@z7I3Oqv~__N$>CII1u7qxshxH> zn#?{4Y8od!4g&73*BGovFIvsSON?qq@WsTVuB;?h2g-M*p_q(ylR68`&vWizd-IA^ zOo1u6P)BL-A_4e=$ZPiT`_T+t8#F*sz7GL&ldIX%_ILEG8pj(lQNBd0*7K}?NNawK zVV&)M=NPQ8wFG!?&ggkrkK0W|F-YnP6s({7^}}!85fUVSz%;B4Bq10(3ZM(d@6jhP z(e?Z+rV7*_D=Ji89}k?kQ+1+~$sP&oxxo@9xU@Eg!!ZwFf2GnB@e8CC&Niy&iJE1{ z?53=RhrxLt9MI#IL0BXx8|CQvW64Ej^IpE3J@3^(Ffi z9|m~3STLHd>`7E`IGriN^DuUpOAe~keXsTedbJtZ{Ur!$>zY^ zFeeQ{NgT?KSJ6txpXNR))5R+d_Ng+}ahk|JPouC<;c|Cyl!N6s2PP+rtna#=*j$xW zRawt)w`-RD_)ak@{;CVWuQcg5_?tDuRq9gai_AeCS~@!1K1yS{oYX_`aI4Dtr0WQd z)OIlyKc=_DBq>_%cwLM5W1yqZEHTg%UCK+s*;;`t!L|sad|q}xtwL32a`JgS!=c;T zJrEVFJ+H+Cuoe@gaqvV|@w1OND8e5&61~wE2pZ80zt=+%Jtz4E9S3#YeWwgAX~giD zwZ~;^CXx9PxlgCwrD^usX-d&;B9+dsFmT!lO0pJ>6-S0%V!t)I{J!762%?v}SZcBU z_N~_@juDD8H@hiUpVSR^f8OxC3KD2@+r)sAg4!`QhQKL>%t#61Fcj6W(n|k8A%iAx ze%yNzCxw*U_MO1VLk>5pJKoL|>|{FM2MSx~l3mIyIg#Y=fE8iHD`Mw-lrk=HtiGB9vsdBiCB zr$Yzs;vfxdd3%8!YM|?tNjcm=^ZxO7qgQyp~#&A|qT~8pN{C(5> zHAB@<*4^4H<}oGRd1(BGFXklG$246ocd|Ab1aF&l%|Qo6P%=m!mnl4^Q6;&h8NS^L z(8do1NnIrB%O9^X7Kj}j*ZJW_=Y_JUW&cd@y~j@t(9dO%5grRrT3bSwc`U9YW25KC zh~YLdyn7#F(maQE@ig>6MJ)=He(Zw?1Cp96fN1}yDg!!HQ`6udWA;tspZtB|5w)E< zf4asE~g8Qv#ccRW0p%e{r~E?K@+4C zN!5{9c4U$H`j@XJZ2RE8n1I6$8ME&Ek{gN61}Q_i&p_v6Lyt1Jnbb=O2ES8xTrVro zmiWEwCyoat+;`f+U92HAed!IvKFRfAEkJuNHG(14+4JG9XcFAjQZO%Mr^qWsSBOBE zzCMZp&znFW2vj1*|5L;HJQ+Rlx5eA@pQ5+b7P%3HDSYOLA!5zb?!vEhXcjn*n!EPW zDqzE(rlyJ7pRn?|`L-&5QVT%E2ouCenP^8E-w<~*Tbb506lu)lS&!0OyvilQz|e9k zOumfm8%y z<8TU`4j5K&@8T+}R~#~(QS@_4VJ+BnTut`RVM7NTNeXD_xo;M;M`9W{QE`}eRr$r3 z-%iqM`Y@7hs;LG$$@T} z)4LY!WY!*RYG-qxrPu9^?)64DEH*9@f8|YQszBBEy!N&Elrh2#)?*}Ts~K&2;husq~Pj%$@rYSE9=*xk5BOWEcOC;U@7=mKvmfq05y&1 zR&(_3#TiiMIYLtOjPyPyF`#bSC0A>PHmRSEA`7?5cZ69#Bv#}o($G2X7ogsoN44$N zvMTcOZXb=IkW^K9Ua{0sf1nVK4~@C2T?Yt+@>0~kN&M-Kj;Hj}`(*<10@A+nS@|cb zJ%E6pdoE8+%fn;oVQkpACeh8B!iCqQ#Tj9RMp$M{z1T&v>eApjNLT(j*Hg>KTsf(# zP~+@Nzp3b5~Uwp z*SazHHFmPO!kvX<9>4llDOC(V>#g6H6D*FT#poW|7_<18pj{(pyS;JuXXRH4J^phA zx}HP5p7Q2h^5JL|_OEDiPJLL2c$b_K)ba@_G+cP>@3I0CX5;eX&>16wY+I~%r(oZz z&krHu>Kc*eB`jC8KMC?JKPtiP?z!9b1m-TiUVMUbb8CTz=Ul`>3ze$;D@?H@6DN!{ zLBHmTyOk`jqBL3kJE4q*QDakLu)$-xS(TIK{0#27YfYaCC-=~-7m@juP<@{9NA`E` zuLxtw1kKPVa4kf2W5w?Yy*D5*s0-GzR$FqWU_{GRNmkpv8b^*zV%x@{>6**RE&{#+ z<;gWkoviPZeu5kyT^})uWO-pa0>wmWLbtNHD)9rWeavgiy868D?pZP3gR`y*k%EZG zDoP*e8-pV+kq_{21*sJG8}fyl8BJ#w|5giAT55oVo9=nH6rdb=i5^#2AGkmMz3dyv z4jw+V+U)KSE}jpC<9dw}%fnM_0^X&VFD@h50uJhpn%PU)qPp@ilKY*gm~V$ua+B70 zvf=m~S63Y8l6{aw>xn`wF*iG8d{SKPvSstLAaLlTU`nhN5Gg7$n9%~nBX!Z}XBM(4Z*M9es2QIwR;n;__| z&Tf~YrIsY<`lHM(cd)|@un;nk$#Bey19-$o<0*|@Z0V+zM2j32@Mi3~5ifRUWbY4? zfyH+dfGLf%EPm(VK?FV3%XAwF%HQ%iSLI!nv<~?7e3|KH=*NzXe)1;QoC-{l8;79{ zabciab!e#>Etv#k11*p?=l*~X1X8L;j1?(XV8OQ$V!QR_C4mE*d?LZbW{R9MIU!L|e8u%0L#(POL`U1tMmoG`9fP=;F0W(h+UQaEOBm@tRN|ry^^qC>elZ~&3 zI;<622+)eP^}|J`U!N{tnQtEh9vLJufmk=v0>W`wQY-_g5>kQb;>hvQAhGkVQ#)Dv z#}ug8{L@p|A9k3aZ|DUpi;LfTpw~x_A&>|ZM^pzL$OJkdhq4$0>M@-JBRhhvM&%AO zU1#WY8Hjw(b$fUqZP@~Ha<86u?}>H0WZy6Ng#Ex0$ajHUGU-5~GMd_QvORHCoI#8Q zGwbX%$+|oN%<$h5jRO+dzc_uO?b`SYxBc~t`RQ%8&^ z%`yC?iONxx%K8>u0uZE%F1BSN#Vx?G!VcUdkff09=`!4&gQ$@s20Ob=8XWK3=?Taa z&56uojB9BLV3jZD2VI7yUWXnD{q% zm#53r_o(8IWP99-WPcQ-1J}v2hl_y-$A$Xgh?DxwO!IzHbb;FYXKofFA;CEFwgH5c;=~j&FJB&DMuOMU~~6|LV+}bLLql zqS$r+D+ED_PEJ^&?TiDl56~R9-MFhypAV<}kWf&W#B)vWwR*k_M;z~Lxow*4KZHpL z5SoO-h!mT>h2&OUwgAtR*h2*|i@g}8h@OY8mE^+>z}yJ#qvmHpU}ZjZc+skmSvb<2 z?R8vH2;`+l{ywfDL#Cv{_I7cFasqS~ zVBGkVs~FjRS|}EtF#{aM0x5Zzf@)OAsvP9^$fiJx^jAP{rF!nWrQE5gTLKHqn7UV9 zd(*8OgKkZB6r(()F|l#7xMfWCtIkQS?wLm=JDebMqsM(@j&F`?v7?UeegGD^ei1Ly zIkFfJ)TuN+hQ`Q+ortP&eb`spKJT=5{g}USuUp&bcC+^VpV|=l%kV^I9$XSL1k+O- zwR@sOk+UYN>?&MAGK_FGY*&{LNHx$%yJ}y@H8TZy7!%XtvB@QxM?k(uoe#-^312px zt}2=E*+Xro7JNq>r=-o_FC@80zgx|h2M15Nw4w6_Rm(O#l#1f2_6nP5^jLr!cj3pu zDLhHwTAxwrwgBgGVtFh~0b((q&WJ{_8n0sDurhC0Tt<@j=l&}k-|g>A5mUd~FSEHy z8Y{gV_4_R)evCsOfeRe!kv0#G>R9H5Smq#7$x^TZN0J<`S8kgT0XncC_=0*IJsscm z^@-HzcEty9cSR}v63q$p9o6E&m!QrqXRDINqYp%s>Vg8({`U$1EOdle_~S4@s$rRZ z?Rq(FC3rHRbmI0zF?HMift`~?(VU8YiO5MneB1luXljLW0KI^fQUg|xOpACzb4m*_ z>s7_KRK83=oG6*U^I6bQ-{joWh-Azj%UX+fI*T+yLx_i$QcV@SZs}A1%f?|2X^69k z?}z9V+Iz@n|i)&BwFzSVH?ALEcV*-)buCcS{*Q zjkyt5f00RZNV_woxKmAK18d473Y^NMR|qccI_US?_HlG@*)MclmacTyx3!yX4%Y$e zt(J}`NFgp$nB&%4?J~!^sjVmTOX`&vO+DcZ6T7u5DOEapPs5UyEd942>CwTW0q;j1 z{9f$Wo1AZdQR7yLl%4!Sc%XC@3jhE|Ik~O|l0~wa+-!8qj945k1o@h>IK(8erwix< z_GCxgL=$t+-3vccYY!#C;mZ2kp}~}Y56t}s{+|PM(xNGX0_k=3{ejI^%dhV}?sUG7 z>?>LB^WheVG8BmmrSjjUg;3C9TrwIxj=U1JgF`oz>HG+>riEq!>m}11p z3UF9lI~PvV|ELXT#wXGpX zS1Smdy3Opsg(jeHLE-uLPe!bt@@Xy@)@4wt10ZdKG zDmrX8Tlr`Blq=Zn>4A*yhNN%Zb#^@3U>ZVQH?h~*a6n(dxPj4BBx zMKR!DO2S-X8Q^AUX=xMjA#BDnmnfiexGv!!z5-oR6G#a9_4c(jN_(9fK0^rwUT+Ew zm_`{gnmtq@CM`AI-`m_EZh-O}`w_ZPrM~1@pF7lKp7D{})KKnV1 zq0~9pPtJ30zfV9@7U`HBj-xM|Z>3GT9eKGe(J*amUbWKzHQPI^PO$9>LckfUS5{O& zz+*wt+Vho*EU^9b;`hvnPfdlicQZn>c5XbHI(NtAM$`@shzsgR#79d%Iqo$$PpQ`T zee#G?A% zdhQXE%nWHMJ%bo7xOLg~^*Y+tHLWkJL#l)AfRQBtY64WMS-%y=Mg`H|ESb#e;m@SK zIA1%>9KJvtxwG>LA}@h+zgso$DJuvFzv|MCr+0fkautreU?PB+LOfZ1efs43J*D4| zD$8>h=j8#R!BDk#dp%vPHBYp+78QZPk)QAZY1Zm>r`dp%(fiscj&t^Op6Qy<)Deq@ zgeA$Vs;c$Wn?MWU*>YmUZaW-j`@BBc*!NYqt==t7wZ4Dm;!d9(+w}D#aM|zlH55<0 z2l&%bzCCbbvP~V5N-)8~OueuEg~rJ=87^YS)NvP{KQN@-*4*t{?66xIb0$&X0z6D{ z<}Q{eR!|9%3%;Hdnv|0$=K{e?=&_w=`DEHJKz7PblaC#Kxj@UPxwlcy3Db@ds)Tk&4Pq?C>Gs>Zwf z`*btU{W9Eupf_)S!tJJVWqsdAbl8W>wPvgJmdhM|uT@$_HMQj`>~u4BjA-B2+Yx)d zF6h!{F8rq)m#04=(MZ1zWl}X)U*8uLcCm650vYl7oXZ|7N!i(XvNIwcyp@xYzq%&V z@DSTKki>YuXe+#T#W8tFF?$2_A4;3TZ!iY-<;hSC&=PK9-@y$D6R`+~h6j6(3>h@v zJ^AD(k-Xn}N22P19#$B1<@s*K`q&H(8x;gM>wQ99V1x`uXYUn7H+pkzwtM{q zsmSsqGb*00HZxCq3u6Ve7F2{zv?o#rV~u(rvfb3e0;MGB4zht)O_U@?pDYUJsPf>w zW$#9`MkQB;R!G=$O+jt6#c9?F&bH9>9W zKL74(w^&10L+Fc@WZk`m>ajR)kj64;zdS@?$0}qSWkV&M-;O16XGgU)S^r@HB4G}@ z?iraA7WVJRUTdc1z1`~%>i+`=9z;GojJsSLdU;dfegC*pKn9Z0nX}wnT)0nG#wjD* zU&vK+=}jima7d!meBD6xa-Gy=@r?L z>h|ud-H)?FNBE6`IWyVqwrcfzTenCE`UceXyKN@ZqN1XPYZ)nyEiI3yvoqN&24cVj z5Ga!6K45rI>QuFqlvGv0*@y^Y^*s)gr{BgV$4!70`#f!JubiBYkMNw+vOB1L9aFB= zViJ+k(()of39*{BfoCNC8o*byfHm}okVetGKpu6}HATg1b*Gc4_4oGmQ#2%AN08(7 zTCO4rh9x01&imbG=O0oFz0#+doSgjoV1N&?bIW^Q?idh?&k@cMGOyYoH+pU2sN^|>Dq@cYG2 zeu&32jW~+m`cJIT(qacfLlv<|O&I72>CVoqlz(-#c59jCEA5c60}-V)6s>tIoYO(Y za3n@(WhF5Dud^|haFsJM6ofBxzAR_E5GxeaF=8f5WeNrPG-X6q5r}cowt4*(-x2w*Yq{E`o_m}%s=sZ4{7@i36zI+R_ z=`>&rxzb;r8^54EJmf4v^>>1gbF)F)fVY%YR364CbN%IwyoB@hoNsrIb9^usXEJ5B z{_b=?8wRg7Wa0f(nLf{Qx@uPtcp;07-t>Cj?0Y7cCJFIftxNKUwg1X7%XxV*%N^ha zHSaN=Ok2^@w-%@J9i#OEGd_>ou66!Fpk==Eatx|$`t)!XuP&d(R(`xKZ3D8Pv@_n# z_Emg*6n8|~MUR2+ddg&4-J|`yS)<$8(AX$_lCNGX6QZlM!r5xMbeE-P7M`xegq6j$ z%WMW=Itw0@lg$cQ1BcPr(GjtDGdi{~bCRv3=p+(~x`Pk208M$d(Vj?_!ySV)WV>lO z>_7Jx8nwgilIFWWf!0s=bm_xTEx(<`xQFS<`5x8ndx4Z+*y_vO1I%T}czBDvY&Lg+Azelb*R z*E4qKZvbE0adZ~s?UZR8dpTP__;ktJ*I~2Sa5%x(1Ey^nSOQvBwzp^0N1M$EyqLdM z>(`5B1I9V~+|^Ut_3lW7WHM^%KRqVLnrvFRT&?G0vwLGW*k{|YbC#&UCtMo~cV}iv zVUIIS=-mH6nR~OReX7wyXRLgNz&wl1i~?=7YDZ}n+Uj?kKrR}B5G`J4aI~Wf$z-$E zWO13%e3N>HA|PDZDG}8LvKA%YU0GR)+4V|e)!tPir@;TX7<3FLvFVJ;M6ZNLER9Nl zUz#aEt+Sh(s*6$DSqAjH_ZwSHGuSAXVnw_R{>n%2{tgObj zf6+;6Z@xB|0)1W|f}hl=2oWVMVPIkD5!g-!V7K%>Q*sfYpcsS_ySk1}qf}CrJDtDQ zPj5{9C&sp-uWS*S#d+MWx(wzDqH#HUS{+p#q#ZOhJxGut2?w$BfXx=9FEWlezx_wL zs?AkZy+88Fj4_6bIFKml05|Y|U%7E#y&L)$|Aya5NX|d_|1%l+_x8@q<-r#i7LcjN Wkvu)p=jJ~*@1(`$#cD(h0{=gHoFJJ1 diff --git a/content/ko/case-studies/huawei/huawei_logo.png b/content/ko/case-studies/huawei/huawei_logo.png deleted file mode 100644 index 94361a27eb5af3a06992dda4aa1b411ab702ee2e..0000000000000000000000000000000000000000 GIT binary patch literal 0 HcmV?d00001 literal 14274 zcmaKT19T=`vu&d`KN(#F8dMA5{+*u!DWga-%+l+;35-C12mn#;(}hR)z089H|x`)_R^ARc~q zdjlgY6K5hr6Eh22UXtt19ugu8V_p(9Rv88vdtnoE3kgq06D3bsWg|~3BTi!yem)`| zcdl;$8xv;(B6k~WTPH4eUXuU7<@&z<=QceF(SL|ITk(?or&H=O@t%&h}jN^lol$bZ*Rac8+HBjGUaD^bAb&OiZ-j60}Ypw$29bw6;#9|3VNkaWZnW zuy?kwvnBcm(ZJBo#hI7n+tdH-f{p!u!`eFi*D!q#7`?lJJv}2G!#`d6m!OQy|6kO` z=D($#oE1&}$KU@au#>Wfy$QXdiIbg+qtSQZOiBMSWzQw-Xky@O=csIFXZ^2HlsC6? zwsSJKvnLW3CIZM97+Ki<)28|l1sNGGNn0mp16v~#NfBO>ZwNXI3u7)3F$NAXP7y&d zQ889VMp1T7Mov)5AAHx!9Q4I{!=8_3`k&rq4f@|IOR) z#(y*4#P*x+j^Dhlm%`}<0%Bp66cJQ*U%%9W^hXn2zB*^HV5aF5K?dg$L`FoG9TXNt zLKp-J&n7Vie|Q(=omPSiCn6FQ1yknf0}2F$gB(|5$xDJH8ld8uQMy)ssi=&sX$&gy zV02n9=jLAJ=APB{Jawzp=@u9>jd@I2eJu=o@dvE%pm|PN5y1RDE||5_y^}v}CM}HE zmXo&|z#2Kw6b(^TC9(TZ007`}Hcl{e1U4GT+ijW_OKEq?q4P3g-hmN_uiYNx-JxVw zmVSHf@A%eO>(XDpN>)~@T9HKnu+U(&B5T&e zNAT!bEc^fFqBr8+?uXDFT5l=d^*cpeuO#L{I==EjBc&g+cn~MpTRV|%iI$d>|wnTbRiMhy_GaLOtSFCO~{PX4yXbP?`& zVMr>uUH|Czb$36ambiCFjN{97F8Dg-B4JZrIMDNTl^`(q%U4sHC?F36fR#XV53v7X zc;QP(2v_;$x=5Eg|>E%^Ac3jxg6YV59@*2M5usK_}8PQ25&I&9J} z=t9?gb!zGlY#aPID(+SiNk_qW%ygf!C<#3f`bcjgBpi9;;QNKH)PTO%jb(ngMyhMVPrF;@e*efhK45;#pw!?9*ZL zva~HE+zAwU+1cbo?m_eA{AYGF>SP9LyV4!-Uc^%K zOL_Y*DFcS#>;T&!gX_RF7)6Wj-@q;l;or(_Zeg&a>ACpa)1jOf!W<`FPm+ET= zq#7CI4XvNS0^^^rq`nUZB>jO4M2p^a{?X~zAJO(51Ry9YE0g-HVykI=5^lHyRjSqY@teU@ z8MWqpAB(OjbvRjB{-CQFsJH@|g@m7T`Xq||VMhdM$@!8Chtjn!dQyl8Ctq1=nS zDk^b{SS^@ZEarjH2iDexxRI+7()d*+7Q0c|3#*Jdaj9 zos;cGeCpTuY=zbpDT;{BC$=r@OkNSjH92oUhwN+qjvl(t0V`t2R})0$ufu7D(Gw$V z4daCO@ooGixoM77g-S6m83xB-6X{NC!&qh`_hT3~T%!04)1a5jg&!O^#FoMXp-E=- z`4v$aEDF%i9H_CA_RKx;mo(IdO{tLW&xe?KVtm>IFYlKI{ug_o1(og0OZXeSXrFeP zN53M-Y6Uw5Xk-FF3Nq27Q1ZN$uid7n&>eA+lkG7+<84A^kOQaOnOe#dhV&Sg2qOcx zibJd_snQvY1RHIE=2-eFYjMv-npDdbC}5k`$tH#g%U93{0af0XIEdV*Y+6n%e;Y18 zxf^J_0PnxE-lxnVUl;4OKYP|E-_CXxi%M_BUn~prqf&0 zTMs4&OAthEN3=+H}{$yF|Jmy)CYOu#NPSY@%?g_X@aMV;!R+y#ubV9$viy9hD%ac_eL0QV;+K5?S;#DmjoH(VdQN6ZcD-|4b=dsOXViGxGkv(AdpmMGkz zjaPW+$Fp(EylGrL$7-nOoQ*^7EQPL3W$F{l4Y)cDY0?-O72S0Eug{?5L< zn$JfeVqAP|jd$_TB5IT1+o;9!4Xvfnn-Yo6lYa-zH47oMsA_0{hmk89PLVE3H^ebH z4W68OH&Rj#zg+zy3~Xe?XOqV=L-?a$T>?c#kzq6cVrRH8gj?nMCIS#2-A|8I&aS%f zaf(Iw{2uO{8VcP3JY6z%6%k!MD89B%`6!?$jyS4TK&U9>&5ge-RWo~IdX&s$)wX_Bcdeah-0jM50O(r2n^%#LL(?I!wD#|+kpjJYVIRjuX6X|}7WI*2i{ zNK2gfzQ3@lme-5PXcW0xD4dl`i#`Ku`B(R7M1}wc(N4{boulB^m~V?CGEdOxkoAOA?m#Ewj_F$#52Hslf*QcYlBW zTkF*h5jPqZlYoMlA@DV+4qGq;^b|ys@R3$GBl$W{>&6VA(3(v=W-UnOoj!N6Ndd~l z;?;FI#qR2FUM=E{7>?0ia0*H{cRkL!(8;LK-q@HI&)iZ zL6_+H;IHV@Km@C^_^KoWxLPiOm?AW&=^o|XSKI14V60%$3fFG6SoZPvXEnHjk3HZ7nOxOPZdx@8UU&$v8FKdbg1a{K<4$_2`;!! zEhVPyY-7PaI^qTSvlda*rM%K~0UW-W_ymQKU>kGTca@ue-5AElx3|8DgbSmxbPJv{ z6y26w34<_oIjXkHsX~SOhG?eEZuJZmPrsUQ8ls45Gs)mm$wyA#Y)Q(UH#0tW;IJ$> zxd{D-iCAFKC8bwE1pt@|={{WTpfN-2oUe!-n3+;>wI#)jh7U_@nybt34LmF6s@-(I zaZfnh@C5K^r%_9)Vt|~4qji8sYv>H$qS1a(zo2UQmh0RH`kV+ATmlPI;gcMyP9U%- zK^TgR2C~ptL9h+RNWhO>D5$vgWFI>t$a*6?ds}956uRuq zGmT4S>XO^KoOAfw_5y~l+20b@|Min(K&N1SP@=^V0ado9Zr6Rq7xb=gvhu|AY)F(1 z9|2!>Df|kH%?xyfCkQ#)Xze{+0qvLi<|8aNvxO;J5OJ8z#Dyz-OahR)tf{VZg^y2! z$WYNiUlglyEO`w42oU#h#i^K3%tT!NN>zoBpb}z8SXG2f7Pch03w~e6WSEW?Txigi z$lu;z2Mo5C&CP0GOi-L`%V7x>hgI`o0afEz&M41$fr47V>K^MnjWBo04w(cye&9>0 zNree+MGk&`g^xk^uo%CED<%BY*it>S`cz}1&P!YDX+@>P$gh8hM91C(?R0>&Kkw$G29jO~tq6-9~U-sRHRr!(`#sfKtGBENytbPcP$nl$w|T$^ft`buh5 z+*-RA(-Y(6hBdemLI=*i0##FzITOD?Irdji3t+s{FOtS@X+6hoRhLK~-!u`u93sZ3|os;Ol-tEVqL z3EkXf_qseHafu(U?Qxmi27u#F9ZE^*atKIb!r7MBE&ZK`#*a2qdFUm53ZfkzX5eXL zE`%+0H*-F{Rmzz`Z#m7=RzL2I!by*?vImF+2pnu&4>|U0ERQBk_{-z_%>ibQ=|+Y? zI3b5W78)no*|%X}aFVte7Q(g2MdEGV;!)EJ=Za)tFbqL(ern>FJ)vba;BL(hL12zM z7Hy0;md;mZ@Qc%K$?*xZE8*e6%9uGPN7yg?YT7Kz4QK5*$LLld{pRa$@R?KW^JS@4 zl&giMRLKIE%OA_8^Tnr9*6$RJh8gJGlbu(p+0y#cI`RnZZL@()#FW%^LTL3F&HB-O z%)P~HC5h?kz*xgY8a@yaBLrq(Q0#Wq8jb#G@5@MvZ-8ig;UZcLX=X$0g;?$ii-}3% zG}boUbq3~7=nu1!!92@)=D^pQsV}55OzrY$jLgldEDEg`#GIJ>+$4P8Icow#jmMGy8j(S%b|WH4wY279!vUe; z=<0KLmkv`leLAA>drpB3%gB%+^ocB(0Ur)v~i4jYq#&`7Qc z*Uq9kD2uynV51;#BrzclBaT^9i>;O`1=U1bk-8dDIFlQ$`OqKU-L5($q|{9ec`@u; z$j`=#I!(KCyZVq$<42mM!DyE4=86X!#oC0s-9d>GZ`KsL?Z~DQRxgJww)lMN9MBSV zfc~)99c?FXYR0|36X(+?dN-eZK)Uo`tGJI}Z;C#TCj|z{OR-_lltczevILNcWAL?f zY{8YutEE)K(lc(hjgOD7#gXXwmzWZZr}pN4J(`SG3*@u%sc z#&)+0#fkBv5m+V-?8%L6zi!9-dZp1UpUbVnrbOe4Z-_8-C@khnJgSxGy?aBJlRsqIPs5R;VfGbcXP~TCp2<3O zI#aRpBycq-u63ieZ9$Uj_x7O?6t?sxG2wXmyEV%0{VhQJvj+Y8N#4Djw#%qftPuMa zQHk)`ad37#j(_nYqmJI*^T||>X9hA*by#gHvbUCNA@)ndE3iK$%2iyg#@NpEzDMAP zb)P^)`>Qk66gCvN%n%IK)ufdbN)My-y&rx2_;!wok!>O$A3fi)2y&-2&B}h*dM&#$C zkK4*Ad{JGD&1QT{Zi>5gR)2*Yac-GUAK(wIv}BM=N2oQpz4pr@i!?k2U}v~>3*Pw@ zhSxBykysE<=cLIA&L1p@#_|B}v>0F$KO9rr4c;KkQuChhmN1iNc-qiO$U|;CHu{0vIK&mt{iM!60V2MA5wy zSTL&=o7cV4i6>cK_3dWV73^o)7S7t2CMl+Ajxcpalr5Q*Q2;(C=LcPzq>Q(dNHo7ftSlFKbw zORF*`^mr`@&JkB^IZ>u7J3qyyj5|=jGli3Yf%Dw0J+ou99WJKN0ky83nNsvZAzj7D zM-KpxaSsE58jA{Xk&BBwq46piy?J7f@lqpU9d^gp72~a(mJ$MTCK^k(nRikHszBs< zWwds{Z)L;5+G<)*p5oErNKvtQOriGH?e9flGK=DgMla^#R6@__^+^qvrY*b`oaWpj z#S`CTKY&9~i_eHhazg)J)r<*dnll=>^NrXApUnGe#}1Hw8|*A}j^tTWTx_@xD!PJL z<#EKR=pvcSak8e!<@Fu%F7iS~eBzirL?dujhAA5qQJrMV?ho&$DN<@P*}2Hb&7Dxh z+=b4UYVht<8;R+ZrGv=%w)aPPan%s3evG}6C>qiYW(}oh1Au`%cqdn*#7p$WcF0P1=#P--BQ}I0`+IfWsTz6IrB=@VFRpkGjtuM z_Lm2&H$3%!*Ov$Uk;2+YTk#73hdz9wq`%?80;_50K}mGxz!Z*N##5vsZQH6{$0j^o zTq?7kpPAd!bc0gJmZn!};7*vPQsa(9FmA#7sWB25Ri$6=FVRnDHOiKllpvoI`U8CV z7HCR{!U4=_PR4&ZM4G3~jLqV@mJdDJkJl)7gDwe##4Uw>9AyfS2jq&-G@0cc-|0T) zI95g_B{?28mkg=F*LocBlf5Y3?s08N#zwQo!6LSUJ36B4fWA{D!>ZuiES*ZFye~;o zW*i1RRC+8#)e+{ZzlBUVzkaN#>FGU>CK>ty)4pX_rYlsgHakH7D6^bCSeBXSxzm+Z zOWMODz8Xa?B|fMG zYZ&XHWPM%nQEkg=K76|PDXycz=@m&wDG}Pj>5zjj!R%ZbhPN7`@27J{UzvJ!(w6wN9J z64WX73Wu5>ii08JrrCDIds>PNFtza{MN3tY$ywrjIR%r@3#=%!Rw&n6(3;1nl9EmO z@eeK0oi8P-Fpzqw76H_yZsnN}>zg3YDDHj?k1ua;b-1d`ZmCyT7gUEwM{#+Zhq+_k z4CRzbEUTgQosc0;xFSrZ%C{=h?t>&N)qAAD@21(M?IhbB!zXO6TxMi+HgRJ5;D%#c zcJDK9G$M=LUGd2;LgiQCw6OlE1QfacwMV)RMt#woLIp7dyf`&J7J|_3q;}muDcmrI z78Al;_HR?;+>K+!IfWF`H;y?N9D1Q9F_>?#Ju|vl4x4J3ab@PU$e?W{r{+V3NFl{{ zya#^NMTs!3-qlUz>Snua3*4Z&kXx8B+fOfoiH9ZwhS+PRd|RiTF?_(EWPgRGV~tp3 zf{oh8qn<(p(}Wvg5KZE`l$i)O-`-}f=!$d`IgT()RS4#2s)5rp+@JKYc%2VjPKax= z;?`FVZ~L`6-;aJ5_kD;1Zy(zDqXo&udXx(nmb;5|-z%SH6AX;wW-f#(5p9O-3}Rdz z#^1tArs9Q0SS5Wib4Tt&lVLyYx_`7^;lf50>)B5BWXk5^pU;DC(rCINEy0^}A13rz zE~7;ul%R4Of7t*g$eAH!{z}-J)IG1aG>u~`-cdqwxlG|Pj23ogmJv7|QLV3UB*cg* zA|`bkZGifn@N;NbyF^>~tC&p0p?`H3dC&JOfs%Jw1k(B_v1i9^If&opzX}^>9emH% zw8DS7C)$)ySQGeuvM{i$2;h(z+#QTjWXQhRpT?!0BVodzujhEaU6(y~4Ke}Cvtv}R6 zP@q9%*})ixJ{@g2!Rf4ru}^L^G`^qY3RKjz$di4MB4Pj!& z-Fc-)>VgQVXrvEO6Cq3#Atj?5X?jE4%VK4p%?GG7nPJ_{ccOP7OMsxLq+IsE+rAz; zeJ*rrU6m3bsLf<5@Vx$fDT_h#OI%8czCe(&&pJu2$m<(7)3FbZ1>$tk~Dq%(lvXAI$Z9ILQuuP-tH=UDeo6_(joO`cGYkbue zf%sJwNlQh8i2RoWJBI6&p=igyH%}ILCJ`tMv%!l=k zN_am)vq2)m(rO)Vxte{?+8_9x@tfc>G8O(%vC2#R9;-BkIACekCi?(k#DVdbAKqNK#3 z%U4xU;B~SbOTRe6&ysAx+OLTwNFE0L1&Oh2v+R1q6{s_Pljr)1>{X=DnA`Q6wAe!O zrTz0nY&;S?qD=&3-#a|b{qVp;a(7J6A6ZV-yIz4ow;ur|!@K3G-WB7TD24e>?2Ea` zP0wuar+(>br4Q?+Igl;6yK0$dpC7vZUQO4af$rs<|iIrM8{OO`j>RV*Iinu;+v?F%ICp+xO zlD^Y8&iz;H<~i!Wb_2w1pc2V?rXp-E*6=JA!KJ~t-RxdZyDpK{xQWX#Fw=Iv=yxsV zALhxqE(8Kl;hh;buiC2Nz;+u5{yL;|CCVcwfRDVbcgeT+~ zf3m6~6$m3Q3qmPn25-5eNDgOn#{p~g8K_JUFmSv@T0L)SpHTkbw}GY4Ce(UEUPdEB zQQCNaF-ffww{P-FbC~S%CmReW-aoe%(;wqxtKG* zuVJsu^U015ViUV3PIk4oT1VEsG8~#);ne`V5H0M9I> zM45;Rcfo)I{lg63AmTu!D9>a>OH@XBMAt1<W!$(S4b)7%k?o-_IM~Bz9LH+W>tv@PH?dPK+B#?uJ z;PI8t;8|fX)lTd*kqE^0$1BV=8hhC7lcSTmf**t&!B(FyI7O(KiLPxD{ERYSbRjiKb#|AsUE8F*s{i(#o zp1iT%D=z^;ICt7%miOsBEuHA`2UfmZVQjK(wov$0jfE}IN93Fa=1jQ^6}Th&kh$^h zx{ZFe#tVk(69Tix!ssYk@-{bbv@23Ovt0x0$zWvEGI-u#@$pldJ`v%EfAeJ1+mSAJ zr9n!ig*(8w#`Wvdv)A%a zT4Kf$lBK#^Q$b;Tv$@4t?hOCd?J=&&j$hDWTmBhv&kUtxA`w*OxYdscBE?K&fHB_K za)k@T+aISf5NjYHwsW%jODpFCeC!X1_aBOb91V|Aw-t!zdp|R$ZlZ`3TUA>*sZvT4 z$6S1Xr zKrX@Nk;M{q-8<-JnQX>Cf3pZ?ngWE9-lsWWB@^OA@2-7`$Ybv&=;$-5Gr1I^#K%Mw zMm4~Lakhi@v&dw)l&}O-CCSV{z(EuzQ~}f@E4924+_&CA-9g{Be=pDCUVnh;zU8t1 z1-$W@`dG*$nE-q9<*uMW(`GF0JMdyC6(tB56c9;j+SF(&6@?Ax+dtU5jCQsUP+Q9u1kndigLVro}^)<)Y@)M zgFBg+K4@wRxui+XpfxrwHz6;6mjEQeg1PtPanu!-j~oar1+c6F5f%)QtbK&Na(zH3us4WO?=+04O#5XQW5Mv0t)YU)?>XgX zHe990VuK!!=7yh$B;0#Sp=eW3aS($)%R|7JLKVRAF@uIM?O6oe!EyraKPVxWp(3cUSyfw7_@K;o>8$b7*YRSU0m z>CPFwu!_N&$=>OU71(Z$DWhVh!?kEIv$sX?njnR+ldG3nwvPky?bJM%?hT8eB;Ju@J z-}_DC31@_2PEnWjI)b71>=k#@TY@t35{t@^_6B*Chg)gf_XF1rlA$4|nKL@zx>*iz zwuMtATg(}3j-~*Kw|fzsZc}7D7;$|emEPxP?qSv|{+s$?zC=AAjmKRbvMGhzCEBx$ zmbhT4v=~9thJ2-@1kJdF^@CD=)K{mpP!ZxS>e$F#f7_WqDIWTZ=aI|OjR=p*0>^Ws^7q9>PAQAfvTCVPJ#yutCDm1YSamrw1qpDn30 z)6@o<;sMaO~#u|&Q8u%+}CG^pX4~$Za5r-=-2OU_Z1ai8*gM$ zt3|g<-h;aDYDCE^xXY`Yrw%A72-Rtx}_*a7QI)`w-BWEKDAP}<0wp`O~v&Vcmb3!=xoecWljxKW5J9gdt|=cgC}&u7WOL@aFek^)e)kw6n|I z8(Zmn=x%bf;$-UAdg>vQ;4z2PUFtH>QE6G3z0l>dVr9!=&lr}w3XTDuhRVVv-m$2-tf#CPM#ra|Uc7{(r2|5;Ybki6Q290In zdaX+1x_gZ>JAAOH?CGI=s$veETvL~BWqG*jhj7f@$C^d~1$cSE!|(Bwl-q#$(Olku zcM3&a(s}6Zc$HB?gb-aTfvJhh+srVbI*%#gg8ysR-1mMA-Mh@S*je5Dp(u=+567CB z>!^|a6>YCSl9YMLuC2yQoKCONm6bo#pWZ%NcNi(M9udSz65d9(Aj2jk+qIr-g+mc! zyxd6=(z#527Pq9HfNLaN`|SoydU8Fv{@Hw4jm8d;hqCH8Od8(yaT{gqmsokAB=GY7 z`uqq4knKkRgF`v%D+I%k?QUVXlSoH5~`i)L`_?Wxm@!-9B+ zWZAVJt+I`9xrGlI+DpzpUT!OCYZsSirkEjosi8MpwJty+V)wi1z8$62(dodH&9l~+ zRH@pteZPHJoz{8NCf920r9$CuKogm-EXqnZiW$hpHjZ~;c3rXl;!#Mfn%;hgq#5X zg~g;9og9@{aoOsy^U8HeRW3{B6fiqKr26Zwz3agt$AfX<_4(rW=V7HIugz^H*Td-q zEQq*Q20(hhD|}GXx_h&ttMj;cxd!C6!oG`8nMS(Rh41`{))@)n1i^D-C2V29d`gV8qnDJ#I98o->UPiHiC=?(#FV0Ucq`YBl~rEF2oX_@}D2($i**$sKFg(dh4Uj@#B!ncYqiXdpajNcZc9P|w%NIPhl| zIM*ryBm$m#gLeo&Qv8u|9^VPDmky&LSkn$rxHyIE{A#jM>6^=*C?NyRMkvY zE7$4v6W`~)w;Leq&9=W6b=_CXzq6XZ$b~~EbC}b32>5oYq|&xoJd@2$MS9+*xj|x5 zSf~J}vke1lC%JdqUe_UH6_t&S6U(%IpLcn*n*$aBqr}le@&Hmw_L1+`%nQ4wa=G`TMM|YGLazp<_N~1`^Y>qx{ra8pb)8F# zi!|P!9zBHN@>ed8YvADE(kVME=M?hje1t?(>_)Hmn0mfro;>T#Rx2$w8@yhR^Zi*6!#C=lS|C1 zi+dN2HhT^pr&njB%3gs#avTMsQ5g1qdqJbcvA$okQem?0r^KcedD*NNb7u?Hpna^~ z&gaUqlrE>L)#||U-o|C zB-!o0IMdPMfcOb&qFD^dKlIlo4pVe2a~q>QP?jv+@3O(hcgSwPV0H`T@AIHIMb1tg zVZB7;7MK7Kxi*)(y@CQ`<$+8l8uYMc-Sy#a$(ed9V8!D8i}|ZPG7?6BKf9!{uuw0r z8%U6dUuGrouLyt?l!i$#Wv2DKrgd1(eH4a<-DbI3ugBM}OOR!1KTWUOXEHrHI=ZBe zj(o(@@?<&(m)&;q5iA54B1Z-ju?rSJLrq&#TMI!+Mi8gxvZOM7F*Z4FTww10S7f4t`V=X`c2q)1SLo0Lt{gcTNuZcb8&+IJMWw-sBsZ*8*YA(ZiA5Mf zC)y=zHpbjc9atsxHJ}tM4;*4YjfaVjm<=F+m|+mb3$ZFY>`C76Hjl!ZJq)kj!Co z`C;GUqFLWB;BFsG6mGR#{jwiLUmi~hbb1CQ4=N9VfCrTx8dY6g{{EO{_x)`EJQ^(b z6GxMHuGc!A{!gV?B6kV4>YbYnF;(8E_YKkNDahXs4K=`!qM~oWkezPVFISz{9g21| zCA8XY{`f$_%I=el)oCCC3|`AL^<`WsP^c6#<1=>JPsqsbduwGU!2P}a)O2*;B}AP6 z$!e#-JfhbKHU?I$@;&j)c$>Ib}TJMb`KiDdhx zaJZ=`sUWNichyaNABRTn9Y9ZRvDkOxRZ=w^(|&jM-@8U>^VXaxAg+#jVd;Q&@-*(O z#B^Cb#*=gX$UW$yUj&A`>u*BLj*;_`zL+rZUDp^)se81ZwrX`c8=IOW{Y%yAq{4JG zcPPJWh5b!kv#1P32CQtJGh{Qr^?X=JUS2zbZ0t)H85!0wyrd+v>@{xkxQjsq%KinA zP+Ww~b|*?nnOzRU*iM%!R3C9vN}2QR<|?umFa@xp&DJLmmXkk+fq@`?y>b&juzNCN z8J}aPve@m2l--L9k?>!IwPvolUQ@GnJWl?|;IacV2R!1r9K^)nu!rJ;+VZLIeY`yd z*@<`RN&srLTCYfJyAtu+-XB&pzUWu?tLlCq6Xk6lw;gafU)KA)A=H19#`@*GX{IkR z^X44rmzts?$Y?3^2hdRE$Lo`n0Y{sE_sNsb*$^gZ0Ff62*1gl+ymaMOqsl~@o(7Ni zDntD`P4&mKQQX`e>3}5b8iMgGco1Fo%d?Sg*Nqx$8}_R81Ek?;wOm9nYkRO#2v8A? zPK)*EY(axLS~SfKWUuW)sf<;RHmn~N|6k}#9}dIA6ni&~84($oslqfp(i8Bg~>+j9qwfgSoD9@(8jcyE!{p+24ssPEL+a&dfjt zx%+;7{#kEsLeTgwg>5!krY0w&hYr+YyXL=~N_5`t&tOO8%lo0kEu$hMY4mJnMKQbE zZ{hiA=V1y<9ejP8oGBV7EOa`)j^Nxw1#Fwj$~(32%YIGgaMNox8XpYDTvzKVDQYU} z>Ly4G3bJwzs1yClK`>7Wq`>EL1+zlzYaIx$<)-e7**B-7SzrHZ_fvf_K W@eBhBOaJ+*PEu4>q*h2j@P7e3cUl|( diff --git a/content/ko/case-studies/huawei/index.html b/content/ko/case-studies/huawei/index.html deleted file mode 100644 index ec1cd212f2..0000000000 --- a/content/ko/case-studies/huawei/index.html +++ /dev/null @@ -1,73 +0,0 @@ ---- -title: Huawei Case Study -case_study_styles: true -cid: caseStudies - -new_case_study_styles: true -heading_background: /images/case-studies/huawei/banner1.jpg -heading_title_logo: /images/huawei_logo.png -subheading: > - Embracing Cloud Native as a User – and a Vendor -case_study_details: - - Company: Huawei - - Location: Shenzhen, China - - Industry: Telecommunications Equipment ---- - -

Challenge

- -

A multinational company that's the largest telecommunications equipment manufacturer in the world, Huawei has more than 180,000 employees. In order to support its fast business development around the globe, Huawei has eight data centers for its internal I.T. department, which have been running 800+ applications in 100K+ VMs to serve these 180,000 users. With the rapid increase of new applications, the cost and efficiency of management and deployment of VM-based apps all became critical challenges for business agility. "It's very much a distributed system so we found that managing all of the tasks in a more consistent way is always a challenge," says Peixin Hou, the company's Chief Software Architect and Community Director for Open Source. "We wanted to move into a more agile and decent practice."

- -

Solution

- -

After deciding to use container technology, Huawei began moving the internal I.T. department's applications to run on Kubernetes. So far, about 30 percent of these applications have been transferred to cloud native.

- -

Impact

- -

"By the end of 2016, Huawei's internal I.T. department managed more than 4,000 nodes with tens of thousands containers using a Kubernetes-based Platform as a Service (PaaS) solution," says Hou. "The global deployment cycles decreased from a week to minutes, and the efficiency of application delivery has been improved 10 fold." For the bottom line, he says, "We also see significant operating expense spending cut, in some circumstances 20-30 percent, which we think is very helpful for our business." Given the results Huawei has had internally – and the demand it is seeing externally – the company has also built the technologies into FusionStage™, the PaaS solution it offers its customers.

- -{{< case-studies/quote author="Peixin Hou, chief software architect and community director for open source" >}} -"If you're a vendor, in order to convince your customer, you should use it yourself. Luckily because Huawei has a lot of employees, we can demonstrate the scale of cloud we can build using this technology." -{{< /case-studies/quote >}} - -

Huawei's Kubernetes journey began with one developer. Over two years ago, one of the engineers employed by the networking and telecommunications giant became interested in Kubernetes, the technology for managing application containers across clusters of hosts, and started contributing to its open source community. As the technology developed and the community grew, he kept telling his managers about it.

- -

And as fate would have it, at the same time, Huawei was looking for a better orchestration system for its internal enterprise I.T. department, which supports every business flow processing. "We have more than 180,000 employees worldwide, and a complicated internal procedure, so probably every week this department needs to develop some new applications," says Peixin Hou, Huawei's Chief Software Architect and Community Director for Open Source. "Very often our I.T. departments need to launch tens of thousands of containers, with tasks running across thousands of nodes across the world. It's very much a distributed system, so we found that managing all of the tasks in a more consistent way is always a challenge."

- -

In the past, Huawei had used virtual machines to encapsulate applications, but "every time when we start a VM," Hou says, "whether because it's a new service or because it was a service that was shut down because of some abnormal node functioning, it takes a lot of time." Huawei turned to containerization, so the timing was right to try Kubernetes. It took a year to adopt that engineer's suggestion – the process "is not overnight," says Hou – but once in use, he says, "Kubernetes basically solved most of our problems. Before, the time of deployment took about a week, now it only takes minutes. The developers are happy. That department is also quite happy."

- -

Hou sees great benefits to the company that come with using this technology: "Kubernetes brings agility, scale-out capability, and DevOps practice to the cloud-based applications," he says. "It provides us with the ability to customize the scheduling architecture, which makes possible the affinity between container tasks that gives greater efficiency. It supports multiple container formats. It has extensive support for various container networking solutions and container storage."

- -{{< case-studies/quote image="/images/case-studies/huawei/banner3.jpg" >}} -"Kubernetes basically solved most of our problems. Before, the time of deployment took about a week, now it only takes minutes. The developers are happy. That department is also quite happy." -{{< /case-studies/quote >}} - -

And not least of all, there's an impact on the bottom line. Says Hou: "We also see significant operating expense spending cut in some circumstances 20-30 percent, which is very helpful for our business."

- -

Pleased with those initial results, and seeing a demand for cloud native technologies from its customers, Huawei doubled down on Kubernetes. In the spring of 2016, the company became not only a user but also a vendor.

- -

"We built the Kubernetes technologies into our solutions," says Hou, referring to Huawei's FusionStage™ PaaS offering. "Our customers, from very big telecommunications operators to banks, love the idea of cloud native. They like Kubernetes technology. But they need to spend a lot of time to decompose their applications to turn them into microservice architecture, and as a solution provider, we help them. We've started to work with some Chinese banks, and we see a lot of interest from our customers like China Mobile and Deutsche Telekom."

- -

"If you're just a user, you're just a user," adds Hou. "But if you're a vendor, in order to even convince your customers, you should use it yourself. Luckily because Huawei has a lot of employees, we can demonstrate the scale of cloud we can build using this technology. We provide customer wisdom." While Huawei has its own private cloud, many of its customers run cross-cloud applications using Huawei's solutions. It's a big selling point that most of the public cloud providers now support Kubernetes. "This makes the cross-cloud transition much easier than with other solutions," says Hou.

- -{{< case-studies/quote image="/images/case-studies/huawei/banner4.jpg" >}} -"Our customers, from very big telecommunications operators to banks, love the idea of cloud native. They like Kubernetes technology. But they need to spend a lot of time to decompose their applications to turn them into microservice architecture, and as a solution provider, we help them." -{{< /case-studies/quote >}} - -

Within Huawei itself, once his team completes the transition of the internal business procedure department to Kubernetes, Hou is looking to convince more departments to move over to the cloud native development cycle and practice. "We have a lot of software developers, so we will provide them with our platform as a service solution, our own product," he says. "We would like to see significant cuts in their iteration cycle."

- -

Having overseen the initial move to Kubernetes at Huawei, Hou has advice for other companies considering the technology: "When you start to design the architecture of your application, think about cloud native, think about microservice architecture from the beginning," he says. "I think you will benefit from that."

- -

But if you already have legacy applications, "start from some microservice-friendly part of those applications first, parts that are relatively easy to be decomposed into simpler pieces and are relatively lightweight," Hou says. "Don't think from day one that within how many days I want to move the whole architecture, or move everything into microservices. Don't put that as a kind of target. You should do it in a gradual manner. And I would say for legacy applications, not every piece would be suitable for microservice architecture. No need to force it."

- -

After all, as enthusiastic as Hou is about Kubernetes at Huawei, he estimates that "in the next 10 years, maybe 80 percent of the workload can be distributed, can be run on the cloud native environments. There's still 20 percent that's not, but it's fine. If we can make 80 percent of our workload really be cloud native, to have agility, it's a much better world at the end of the day."

- -{{< case-studies/quote >}} -"In the next 10 years, maybe 80 percent of the workload can be distributed, can be run on the cloud native environments. There's still 20 percent that's not, but it's fine. If we can make 80 percent of our workload really be cloud native, to have agility, it's a much better world at the end of the day." -{{< /case-studies/quote >}} - -

In the nearer future, Hou is looking forward to new features that are being developed around Kubernetes, not least of all the ones that Huawei is contributing to. Huawei engineers have worked on the federation feature (which puts multiple Kubernetes clusters in a single framework to be managed seamlessly), scheduling, container networking and storage, and a just-announced technology called Container Ops, which is a DevOps pipeline engine. "This will put every DevOps job into a container," he explains. "And then this container mechanism is running using Kubernetes, but is also used to test Kubernetes. With that mechanism, we can make the containerized DevOps jobs be created, shared and managed much more easily than before."

- -

Still, Hou sees this technology as only halfway to its full potential. First and foremost, he'd like to expand the scale it can orchestrate, which is important for supersized companies like Huawei – as well as some of its customers.

- -

Hou proudly notes that two years after that first Huawei engineer became a contributor to and evangelist for Kubernetes, Huawei is now a top contributor to the community. "We've learned that the more you contribute to the community," he says, "the more you get back."

diff --git a/content/ko/case-studies/ibm/ibm_featured_logo.png b/content/ko/case-studies/ibm/ibm_featured_logo.png deleted file mode 100644 index adb07a8cdf58888b9884161d64112179b2987112..0000000000000000000000000000000000000000 GIT binary patch literal 0 HcmV?d00001 literal 11298 zcmeHtWmFv9wk;6c2?2t8aA>?~+=IJoBZ0=DaR@HKLx3Q`o#5{765J)YLvRTG$anPJ zbMO88{=6QeyQ+7sIrox1t439=Fl9w)3{)ai7#J7~nfDT^FWb47uP-v<%da{$U|Msk>%l@BXHY$q0Oq{I+sk9Z8Da7ra%qh58xmW>ILZ}oJ6ar3WA3&-SQh$fP z{1c?Ibar+Cv9Y$O;Q4dpZ=U~^*uTXq z@KPO+xRW{9+1^Rr-rhz?^iTOHlr8O@?V*E+ z7vly3{$l;Vq5ny1?gVi)H2EsyYxG5^LZ~kx{6q9YsOFGn2n-Be zql|>8x;yOgM~}~D(jL278Y?pfA6GdAR|In!xeB;)CL)6*m;#EX}SIK@+l-?y&wMuLpEnieC*jaf5c0tv#E(?tYcsU!EkE zpI;TvY0PM>l*@8|=2J5$Ki@vvX1f^Pyl)%1qo43dY#g>76w1^u;2C7?xDwGnuDtqx>q?4M{I3+V6CY>Ir-z@8D1K>r zx~Gf%{5Rg3!2X;)i^pU6FdxOln6eGv!^Z(rI}mIC&CS4xuSo&PPqM02pTp;|VIv=F zv3C|GlS`{K3(SU%QxomS+U^d!rK{{E7is)un+Et}u)ZQG-@cjY8C4$AlpA>2okRTH7pf8tLV9E95UZecwga3>qd$`)d16kU8sAGJt}zn*G$0U|~D1ckk; zdW~YS{fZZT*!Oo&QjPn#toC(*VZ=cB^DPsV0$+{4pJaBv%+yA{!t?AZrfxFKew4)2 z7uoHtINq2cf?p&!zJch3WF7f(22Wm2Jd|whO+m8>3C2v_x|UWY-X5Q!r{nJBLR;dS zX_)0s!$>q4_`hGBO*ei*N0j_Z%>ZAU6ajwsVe`E25Y5Slk%~oJbP&&!H39fEAGFp; z@rsm!yzh3DFtn58M7z(1m4a|9SzPoZg#e7*0W$ZXZ9mdB=Lvai>8l?c(R-{Wnho6xk22`$5yk)i!z&;2Dpncc`tq}24{ z1dYFl)S_-!D%S5(s(!kt0x65D<3%?Rdiz}8)gaZml~fJ}d}`4r_-`LyTixWaBj%!E zaP+fY9Xs$vuttkaIOzZKHP(N+=~i>X#@Ld}SY((Z*~2E-4_2r0#Lb6gp*NND!X0`| z-%$;4aA|)qzB9Ad)ZEEA$HGz?KKW=Pugklj+Gp|l{Lx~Ec7BN}ZN=w|x>cVn@SuMYE*CH1$2Ta|lAg1Y|avc@2B}jLNP)&tm+eXd&jooWCY+sSDvRn=kq{V3p z{mD7WI;YP^&kqQ|Ft(}-$r;W+dX$S6ArZXQLAKc14oiAiXtf_AI9X)qcdadgC{^me zqTh0rr?wa1p@@T`?4NwEV01AuIdiSdWzP=JouuNtK+%cJqlp~~XdH^kmBbzL%6|U5 zuw8T1bNTQ+FfHjd2UOvGW71(eb?NX#WRGL*4PK%v!{S_wMh149rvb2yWO#ripU}BU z+O$_=f)l1;VGC+zX9JXqs_V(-?9WvIoa_ru)j~#{q{(vpxDBOj=ZC-0%PN#PhY8pD7Lv>*uNHG7Q|wM9F)fs&A=8A z8w2+DnnRK@8`Gq{*Gd+IGBV!?bLGu4lD1;^dKoN{;fHO%53DqigWM)8^>6Lqwoc;L zIOO=)nNXsnt?1Fq=`WH75M#+GN4cWF#Thl)%KvJxgRyQtP!;A(`~2**+91zHliOU- z>$L+%B2%)fS`_Jh+Gp!D=Y@ETSqPGC#WLXC@mzP-emcvN!TBUFS?f|9>uyAoJD~AP z4+o&#zU+ML_d|axXLl@U8|m5E@Lhg36*f^}f|tUlQwkL^_^g1DTr;7RI9w!99?L3c z!GvChDAIXrhWyYpLOepm?IIA{k(F^6TS~vOUGCb>Afc@+kzDc^OF)Rt&G(SxfV5Q< z38YcZ&nfFoPyY&f72t8=d1K6Se|#0Lf7K_w@;=Jw(y^W@VPdW`h6( zl&1X5RMa~$%TKnV5PM0baH%GU^t)3ELSE7%YWnP$Yu@2F`;q}-?5UWY0g~q?X+5N>~cg>qH5c_7_ za!3}bRGnyGCMrb^Agk|DNF1(2PUMmRjbMZ|%@IH|-}Zl8LCDtO86edYciI!laK`L0 z%WXGTdqr>?2AN&Vt4s%S6KZ7JIFDOdN#SO|783fbE5}rU_D^*Av)v_)h@j)V6?Obb zOvJjn>e&S}n;G6rjD)8Tox{!MW)*l}8FB~ocBRePx=`stT-LMFNQE;zSy^qm&myfN zZ60c!<2J9i$J;p3M{Fd0kQ7B?0{VI5n+du4{*HKnE`hUI~JQI5hYb1tOE znRCi`XP}4@fn@1!cyko8xFsgrRarBX=*S$V#Mk7aGZ(TRpVSN=;+ZmYvxuFLt+MKX z(XFrk>F3RSWGaZD`)95QVRL(G?l=G!F#T|OOzpJwWu{fvP=vSn%hwY@cDAvWDjLDo zij+B*U$Utf@BC}^sJ_M>mL0`m8O>Rg2kVQ@J>*6o9LcM;-F0~6soQHd6;mnnRVKb` z`TVtVo#%E23ti4vI<~^vDJxWo>?2sm=W?s026l>KmL_jVjjO3U?g0n0z}R#zZpnV0 z>IAf`O?njz^+Rxj2vmyi`l%vCPC;;c;Wyj6WM@zG^>zk76 zL?Mk`^O-kp!az1UBSUOq{2xNP8kQ{v${(`0M!@q8*OvAAm^+Lmm|AO$zKSDs+x)^Q z0jxHv58~s237P(>@tZX z`l@w>zx*c^x$o-dJT@l;rt7#|;osAd%O6*D&cwwItwZj>`jxZGbL{PGHK&RVy=~&FdF>{*21)kco)bjt zAG-mGaNRMitq7_D-ns#RXNHLi>ZEWKiAh$xuXOHCLkZ@@L1z5$AlZqIpPbyOS7d#ji^}GCd9binv~-=&NHM$+MQ6&L3H)mRsPDOkv@4_ zf@BPH$x3~v_RPUFmL5G+d%rX%AtS>0z>6h6U;?5Pl}~6qOns+L>>}hV)wQ>&`uPE| zm>X-sGM}v1`t443gSRt9(3PSXs1X*MTB|68G~`oawrhBPF=RTy7+;JQvqP#B>8=7? zvM-n zh9@%{DFPA~hha}&?-e`}i3r4y-_Hf1en;y-n=0DVzVlbKzJ{*5Y2LuO)6_3odiz?> z)O|@*uE}te=`tG(i@Y9CS0xyVEeo5&x+YXdt3A#D*ar?>?I6e}YBa7@hCQgyz7M^v zHE|Wm(I*0TA7=j%9|3}H$ooJMJY@H!W3AXq6f5lCI?|cvjq*{kYu+E(1FEqzr!s}| zmzPt(d>RMc&LzE-!K-Mmi<)))KM98HGsM6aPHT`&vM;gVhLtyz<=R2sTA0HZkK6J= z3`O}GqztgscKQiZ@r3~1jfAA#XMzPi(^-{@UnzR?8sj*4nbdetE#Ib3foGwO-rD#F zuRR1rcxSy16=n$ARP?(6T5uN!Cy&msNW(pp;Bz?`RDDem_Qb4oH0V9Q-P?FM%9N`~ z7Qas4^R6!~PVWNL!Wrk66ALQk^%!N-!|UF?*3%tTWqcLGw^z6El=?6l+Lv4YQDH56 zAK&wCYjCdXd`nIaNQgpbxUPn(rU)JYjNhAJ5#&AsCHjy$Y38&MqT5ALU7{fG5ryx{ zC>Nrs543$(eY_P%O^M}+;T3#1QaP!7l!2m3jN{eX5_Q?pgo^CWVH>fxRO5n}*YMjb z<8*fY6HxtVBlI$G5);YrkkPW2tYTaI9F!s0LiB3heJ9g;FpGnYkPor7KeGT$|5xnY zV*TnDav=~U?E3{-ZmDl?mB{Dg#9Lze;*#ZLth?CUK00Zsd{m`fc-?yMBS35PY*dgi zzltUIMODwn=`E%BeBH|(4PqaxsS&aqW~kd;2M!4&RWXXk>~nI>WgS_9lCOjI&w$Y@X?{FzX+@}3D9G4|sR64D>u*qA%8ZvKo_+&^qY z*RyNVaF_lZ^3mq)4%P3PWThLLeoPb}8d25Aho(fA^&`91dW7V0 zska{6HXk|Atv|8(ZenP6`gP)rp>}@uC;-8<(@-ovpUFiyZM=_p*t(Sr0|)JQgh)oa z^EqX9pIN+pXRhFW`?J`gh7?q0|Fp)UqZ2J>kxaTiP1mP_r7}GznZNU;56r4=Lx;I< z)7zG-xEDk3WX?ZsyQHHrU7tJATHJS1D8fe8SO*b(I@)f`pSmp<+SBKfbVYv59;v6s zBuC5E&6(vGDQe2{emp|z+rkoVBlTn=p>C*#(v??j>39qfE`?1xZrW9E+F7V{Rh+bPCfi-#cvJnVrhDG*vh~Btr z;zZ4p%CluQNGyzH)v=4*m~d!Blwt>Ni;Bwab5AMQ`Jo%CO4G1vc^|1u_lz}%%L6rT zjsr1F#YWd{#B0ov$u8fQgbfg2Enq)$Jstgg9WYv^vrj??Mh&r2hC{^@kFA!)&3GHi zRXutsDuPh7>s!(AMl*I0%hq1)wiOi@fU4)NT82Oe$JvsGz)xZlpq8RS(CmOOtMzRH zsxgO1lBLDecRmele!SIehf3URRcafi)9A}k4J`j~OORk9$WJ*3vX^ohwiziqCq{KA z4v6fk@Yuw{?Ym1Fa_4Gw%cF6F@?e;^%Qro44aQn##|;8UcvH;s&QjAv7l37nWf|6) z4fBF87w-A{USmR~70d#vQB-K)NW~kR;w$)kYC}-|@McyVHYXHc!#*)f5y=|Od}mX1 zqOdA+@0-b~EMU8hp*1GT1xGp!JMM>Z=iP>)ACLU=QBCC&m4SDG#HG34PP~Q^Kd-oS zGPBkBcW_b!Qu)IwLE<*t=m46DPbprUW|7LPYcn{!;J&dy3VVTBCev{*_kH*LM`!_+ z6TX{KZSj?UElX<6BztMx( z9DRh&5CRfRizhHNg1z`a1K;{K*=DA6qxUxs{i-MBB``W?yY9L}+eMh3Co2NI4O>csw2~hRbLXhK@VeCG#|WrkO;l3(&nk^Pr}PQHWJkl zN&o3*4{xbo{!g|}l&lJY^QS}Wl}Ul}3>dIeze>~2L)#ri>>8`?vnwYp-|m>S_Og4k zfCCNC5OeU=rP$NZ&tj-Z3X~q=w>nJlH2le>M*#7b!wsMuG(o${I&D!7Vob4Eclhv` zdvcP*x1iK=Z=VeG8@rZz#Y}X$4ko{pUmZ{e6@l(ETv+IoW6yGPsbnwbVtDB7y3!p1b_m^jjr8iY znWgLTm*;y5zT-c#@ZdYNf55D2CSt3sRIV=&RQUY1u3ROS$r;5X{V}hr-=W|1ft#gi zG4*NG;xT>o(n4U@>$v~MLpmbpdK8A4;PrV&;4WD9P5kpUadVav3=gqm&!^PhLz6Rx zUVVmlG10+1!QaQalx7FnOspSy$1obTQeG<3_@HGUAW zO$Gzbaa;^_>{c+x4Y8dU@FHP&SdHIFYDs-LWtM6lt;1<2mKy8*s2{MeBD5?XlBB zw5y(x^Aar2Eeiv!;Bv8 zHj8d-tK7;RT?H+Kf#(VB9%VWMlM$HvFi|0d z(W{RU@r81Pq5>A4=)(`mMLzBt&xLa|t!oQ?^s@#lH@0D|Z;nSQ3v$qunxEHRA=~sr zf>KsQdXjx4D$}0ujzdw#Aem^Gn$qIB7OSSvs=cCo#qxRH3wjOwQO3*0;NHF@)qO_{ zX`5wdPQil-`Wjtwi}4uv_uu%_Om2C?jk*B@L89lEMTH|`^<7i((;@tIhQPy$h=PTOJg9Fc>a0T&r4`ym)h0{XP51DRahGWu`VxvXVW%9cZ zLcP1i^khWVJ|nW!w_AMkzJ)f&JC)8`-y)6z0OzHoi zcCOdnA}8c=I2#_7Aaf@8^6YVCl`zp5S&O%m@ux}IOjV)&Y3SEondweC;Rb`JHEPq> z*$FM(3M~0sN^PF9niLMG;D}F6{>!rHG&|h~^vhoY?!wj#E5`zNDzjK7^Qg2@&bf+D z1o+sD0)-r3Q#~mH)((U?(>~#So$N}`IcQwmzVUG>Uv|z_3)>>i-&@z*Hj_{WI9nfbx%i?3_Iap-VLWlIx;e-#Jq{7<9C{y<36v) zcYiRi6X_k9sJVsU)iLmMbC5!G1E{W0Y+!GH|Ij;jOAr`gp)?S)5_G@%0sL`Uq4AhZCAkUSelq3`?Y*_YQ0BF5{nM5tod< z+lflvRe55dkHu}!E6z~xX$}8yiVSUzC`{zr+$_29;!;v;zEF?;!WgkeTo_|p%u-$ znvKu>aaBP$&3H&Icj0@4jd7shoIZ0LlgHvF_8&nE3B@+dn1spK-$pn`1^Bd_`Q05i zX*;u%-$*}1401=-e5R}Q*&XCd?t0@FfJ%m`mK&!r?*vxpYq}ypG@rw_9~(~+-~33j z+mj(g(D|Wb@*@^8I>Kv0z+7jVfz#%8QhF1SN>NCMt21UIK){c4nU5v6tED9{mTA_~ zwgKaUhu##4=>u!V1FkD-`{`X37pxi)Ji#W32EIILL+dl@GWAF9OBq?(9jfrikI%wm z*;#>&4t<^e2}%k)XHfGCV`v!dUEY}-8}9)N}VWqD*RS9DH=FAaBUoud1drR+8Y zJQB~c=uFnzM));fDYy7_NqF9w>u>Hi&{k5IBmoS#TCvQ_mi|qW(sTj?iroLRJ16Y?g?T~!|T=~pum684jkmH9sctBNxruOJ%WnW=IjF?L!TIJ9lDJ*hs5 z9b0IsBzufYH~cKWz?i7P_1aL}B$uXlZS)jgvC3D73P>e4qlhDp9v;T40zqnNHdZ2? z*B8~3xlGh!En-PS4S!?Ixz7RR{_PRBoI<;$!|+GLU56-TTn7vX+mXl_@_c`8nw`kd z`ueW|wc(N(22x%fjy4SR^9_Av@R(?uV&1nAASKyv(v%a+eM}KVJu^{_ZTHiuljOG! zsGplK(%)}01!)7xiUabo54_cLTb(EwrEjhV?8V2NoD*GUFfQCrlldBVn|HS!Y~(?X z{9zkb($VRY>+k0twH1Vb>*x`8B@g{$M+tf{HzrXQr6?KwX$HP`GrVSq8%M53{!BZJ zLr4^if&@Kh~vQoIbS_B64shJ)bRT}fVINiw+eE{m-jAH~+BhgV)0 zRdRpuLM8QxIM&}?RcB7G^4u4vP#5%8Ep{zChr(bt#v3tVt2+)tfX~S>P8<+lv5>w~ zRe%*pXeiXo3{~&>3PdRWBA~tYb~3x{P{1`H@o*pHV0BU{NXika^6ST$IV2R5=~E-` zl_6S1()+)}`uxH#S=& z#}2*WvR@7vbKrjETK9zz(csInDhnW8+x&HM+bMb#Z37lurt6 zF>x#KzN^Ezyk>y(4QLMP8}*mp@Xm=I#lHIOeGr&q)tU!c+xW;fYn;(;8amRcnj6gH zc4wZj8ks0YW^=?#a-tUC?y(ib=bq05-Vz<|Z=OzlsynIhsYii^ zf^!OQ$i+{3mR9&o@Nh-kH7N4My&JK@HMd%##u?ait3Z)MnXf@2<`}7E+FGsgms+vs zXWzqeF2D~?XTKHMHGQ&1<6`D1QgiU9j2W4daFl^2hiRX~I5WHfOu{$(X1RL<_fzhP z{aR|mwdtE8-B6`@J{1(jM;ezvTi|;A7qtKezf6%3x?`(i_k`k6qW#J3cFkl&vc)k} z|5XoRN&dH5{HR)rU6!|SQa*X}eJ9v^{K(XYwl2WEe7li6G@CXIxGOPZlD<(aq3S{# zUJny{{>a1y|I`4-Yc_)^x}`fSen&l~_EN1u1tD6TwZh&QM`CHed|g~9gb&el)3#zG zqbK&KSqTNrRVeu6_lCR8W|-F7>bjCc3KNtm0?e_k4_>+hI3(m=6uYNo#CVU#1jt$k zX-h_VO2ro>qqO}GL~b^nnTxK&nWl&!Ue>~d+xSuHjPlWpJK%i+f6N1#PZ0yk<~fqk ztyZ|ZF@gZ6>j5a)k*2fI` ztH>3?J~$$eR(Ws6LI9rnh{}XE6pWsrLmbZKLe=-pI_OAaG@is*N=&{#p)Ho0#wt{Zj&MHa%YHO8m-}DIu zz|FFG(qdQaeUP6OZdYYScRK);JLh%Fh;0SV@5BfGN`D3l(rh_n^1sdzrc=svgR#VV?0$WeCaQVfmc#tb|~o<`yi6+>nE z+Bp02?)_>YZ+2|~g+vF<=haxdN+VV#j) zV>PIX$HOfoxdA)wDND0f`u%plFo;x$|AKAXh&B2!4W$33HUH;$Cs|4kQ$m>-4u`~$ zUg**m@5wS{y`i8o1)-U$U4I=AqjfYUBR$qbpK31nFfkrFsStx|X(1jV4(t>cxrq)L z9}j&as2Py=>{N0T(y_;&ukDWA9K=)`ow+>#ye@dWfL4z~v&bCi3tKbfru4p(u=*Y} z;!`?q+2$mk?1{AcyXIEDvWiF?qs^?B8(KFYtwL+g;``!QaBIrOCY~7#Ztr`EeLGZw z0U1Z2zD93~D=IrD2)^u7J+*{njMZXT3*7W0FnVO>N=(BKD+7PLM~~!i;|-}0_f>Xo z25o*}?k^7t=B@Gh_Mu6oMX?wVsy> z>tsUPpi>pyuS3h&;cMqjj=C<$kce9`$>K^N1GpI7|6Mf9mBZ`;cEbuzJ$qfChdX|+ zD`4ucW|szyzVRusOpQuYLN74CM{nK=(1;wL$ox8eA?&LO>u)G0mE>V_T^$~mD!*M= zC`-feAsYFLe{*;#CyE?A)9HuZdc56r^2K0Ggz5;`q%7OiM)>A*$2RH@uuJtOvhi%x zGx#miYp2nlQ&T<--uTMicazM9GBn;=5{wG!%=qGC@2gL?rz= zY_G{oJVWueiQpS40}+|%t3&Yibm_featured_logo \ No newline at end of file diff --git a/content/ko/case-studies/ibm/index.html b/content/ko/case-studies/ibm/index.html deleted file mode 100644 index aa3f108e1c..0000000000 --- a/content/ko/case-studies/ibm/index.html +++ /dev/null @@ -1,80 +0,0 @@ ---- -title: IBM Case Study -linkTitle: IBM -case_study_styles: true -cid: caseStudies -logo: ibm_featured_logo.svg -featured: false - -new_case_study_styles: true -heading_background: /images/case-studies/ibm/banner1.jpg -heading_title_logo: /images/ibm_logo.png -subheading: > - Building an Image Trust Service on Kubernetes with Notary and TUF -case_study_details: - - Company: IBM - - Location: Armonk, New York - - Industry: Cloud Computing ---- - -

Challenge

- -

IBM Cloud offers public, private, and hybrid cloud functionality across a diverse set of runtimes from its OpenWhisk-based function as a service (FaaS) offering, managed Kubernetes and containers, to Cloud Foundry platform as a service (PaaS). These runtimes are combined with the power of the company's enterprise technologies, such as MQ and DB2, its modern artificial intelligence (AI) Watson, and data analytics services. Users of IBM Cloud can exploit capabilities from more than 170 different cloud native services in its catalog, including capabilities such as IBM's Weather Company API and data services. In the later part of 2017, the IBM Cloud Container Registry team wanted to build out an image trust service.

- -

Solution

- -

The work on this new service culminated with its public availability in the IBM Cloud in February 2018. The image trust service, called Portieris, is fully based on the Cloud Native Computing Foundation (CNCF) open source project Notary, according to Michael Hough, a software developer with the IBM Cloud Container Registry team. Portieris is a Kubernetes admission controller for enforcing content trust. Users can create image security policies for each Kubernetes namespace, or at the cluster level, and enforce different levels of trust for different images. Portieris is a key part of IBM's trust story, since it makes it possible for users to consume the company's Notary offering from within their IKS clusters. The offering is that Notary server runs in IBM's cloud, and then Portieris runs inside the IKS cluster. This enables users to be able to have their IKS cluster verify that the image they're loading containers from contains exactly what they expect it to, and Portieris is what allows an IKS cluster to apply that verification.

- -

Impact

- -

IBM's intention in offering a managed Kubernetes container service and image registry is to provide a fully secure end-to-end platform for its enterprise customers. "Image signing is one key part of that offering, and our container registry team saw Notary as the de facto way to implement that capability in the current Docker and container ecosystem," Hough says. The company had not been offering image signing before, and Notary is the tool it used to implement that capability. "We had a multi-tenant Docker Registry with private image hosting," Hough says. "The Docker Registry uses hashes to ensure that image content is correct, and data is encrypted both in flight and at rest. But it does not provide any guarantees of who pushed an image. We used Notary to enable users to sign images in their private registry namespaces if they so choose."

- -{{< case-studies/quote author="Michael Hough, a software developer with the IBM Container Registry team" >}} -"We see CNCF as a safe haven for cloud native open source, providing stability, longevity, and expected maintenance for member projects—no matter the originating vendor or project." -{{< /case-studies/quote >}} - -{{< case-studies/lead >}} -Docker had already created the Notary project as an implementation of The Update Framework (TUF), and this implementation of TUF provided the capabilities for Docker Content Trust. -{{< /case-studies/lead >}} - -

"After contribution to CNCF of both TUF and Notary, we perceived that it was becoming the de facto standard for image signing in the container ecosystem", says Michael Hough, a software developer with the IBM Cloud Container Registry team.

- -

The key reason for selecting Notary was that it was already compatible with the existing authentication stack IBM's container registry was using. So was the design of TUF, which does not require the registry team to have to enter the business of key management. Both of these were "attractive design decisions that confirmed our choice of Notary," he says.

- -

The introduction of Notary to implement image signing capability in IBM Cloud encourages increased security across IBM's cloud platform, "where we expect it will include both the signing of official IBM images as well as expected use by security-conscious enterprise customers," Hough says. "When combined with security policy implementations, we expect an increased use of deployment policies in CI/CD pipelines that allow for fine-grained control of service deployment based on image signers."

- -

The availability of image signing "is a huge benefit to security-conscious customers who require this level of image provenance and security," Hough says. "With our IBM Cloud Kubernetes as-a-service offering and the admission controller we have made available, it allows both IBM services as well as customers of the IBM public cloud to use security policies to control service deployment."

- -{{< case-studies/quote - image="/images/case-studies/ibm/banner3.jpg" - author="Michael Hough, a software developer with the IBM Cloud Container Registry team" ->}} -"Image signing is one key part of our Kubernetes container service offering, and our container registry team saw Notary as the de facto way to implement that capability in the current Docker and container ecosystem" -{{< /case-studies/quote >}} - -

Now that the Notary-implemented service is generally available in IBM's public cloud as a component of its existing IBM Cloud Container Registry, it is deployed as a highly available service across five IBM Cloud regions. This high-availability deployment has three instances across two zones in each of the five regions, load balanced with failover support. "We have also deployed it with end-to-end TLS support through to our back-end IBM Cloudant persistence storage service," Hough says.

- -

The IBM team has created and open sourced a Kubernetes admission controller called Portieris, which uses Notary signing information combined with customer-defined security policies to control image deployment into their cluster. "We are hoping to drive adoption of Portieris through its use of our Notary offering," Hough says.

- -

IBM has been a key player in the creation and support of open source foundations, including CNCF. Todd Moore, IBM's vice president of Open Technology, is the current CNCF governing board chair and a number of IBMers are active across many of the CNCF member projects.

- -{{< case-studies/quote - image="/images/case-studies/ibm/banner4.jpg" - author="Michael Hough, a software developer with the IBM Cloud Container Registry team" ->}} -"With our IBM Cloud Kubernetes as-a-service offering and the admission controller we have made available, it allows both IBM services as well as customers of the IBM public cloud to use security policies to control service deployment." -{{< /case-studies/quote >}} - -

"Given that, we see CNCF as a safe haven for cloud native open source, providing stability, longevity, and expected maintenance for member projects—no matter the originating vendor or project," Hough says. Because the entire cloud native world is a fast-moving area with many competing vendors and solutions, "we see the CNCF model as an arbiter of openness and fair play across the ecosystem," he says.

- -

With both TUF and Notary as part of CNCF, IBM expects there to be standardization around these capabilities beyond just de facto standards for signing and provenance. IBM has determined to not simply consume Notary, but also to contribute to the open source project where applicable. "IBMers have contributed a CouchDB backend to support our use of IBM Cloudant as the persistent store; and are working on generalization of the pkcs11 provider, allowing support of other security hardware devices beyond Yubikey," Hough says.

- -{{< case-studies/quote author="Michael Hough, a software developer with the IBM Cloud Container Registry team" >}} -"There are new projects addressing these challenges, including within CNCF. We will definitely be following these advancements with interest. We found the Notary community to be an active and friendly community open to changes, such as our addition of a CouchDB backend for persistent storage." -{{< /case-studies/quote >}} - -

The company has used other CNCF projects containerd, Envoy, Prometheus, gRPC, and CNI, and is looking into SPIFFE and SPIRE as well for potential future use.

- -

What advice does Hough have for other companies that are looking to deploy Notary or a cloud native infrastructure?

- -

"While this is true for many areas of cloud native infrastructure software, we found that a high-availability, multi-region deployment of Notary requires a solid implementation to handle certificate management and rotation," he says. "There are new projects addressing these challenges, including within CNCF. We will definitely be following these advancements with interest. We found the Notary community to be an active and friendly community open to changes, such as our addition of a CouchDB backend for persistent storage."

diff --git a/content/ko/case-studies/ing/index.html b/content/ko/case-studies/ing/index.html deleted file mode 100644 index 037ba9775d..0000000000 --- a/content/ko/case-studies/ing/index.html +++ /dev/null @@ -1,78 +0,0 @@ ---- -title: ING Case Study -linkTitle: ING -case_study_styles: true -cid: caseStudies -weight: 50 -featured: true -quote: > - The big cloud native promise to our business is the ability to go from idea to production within 48 hours. We are some years away from this, but that's quite feasible to us. - -new_case_study_styles: true -heading_background: /images/case-studies/ing/banner1.jpg -heading_title_logo: /images/ing_logo.png -subheading: > - Driving Banking Innovation with Cloud Native -case_study_details: - - Company: ING - - Location: Amsterdam, Netherlands - - Industry: Finance ---- - -

Challenge

- -

After undergoing an agile transformation, ING realized it needed a standardized platform to support the work their developers were doing. "Our DevOps teams got empowered to be autonomous," says Infrastructure Architect Thijs Ebbers. "It has benefits; you get all kinds of ideas. But a lot of teams are going to devise the same wheel. Teams started tinkering with Docker, Docker Swarm, Kubernetes, Mesos. Well, it's not really useful for a company to have one hundred wheels, instead of one good wheel.

- -

Solution

- -

Using Kubernetes for container orchestration and Docker for containerization, the ING team began building an internal public cloud for its CI/CD pipeline and green-field applications. The pipeline, which has been built on Mesos Marathon, will be migrated onto Kubernetes. The bank-account management app Yolt in the U.K. (and soon France and Italy) market already is live hosted on a Kubernetes framework. At least two greenfield projects currently on the Kubernetes framework will be going into production later this year. By the end of 2018, the company plans to have converted a number of APIs used in the banking customer experience to cloud native APIs and host these on the Kubernetes-based platform.

- -

Impact

- -

"Cloud native technologies are helping our speed, from getting an application to test to acceptance to production," says Infrastructure Architect Onno Van der Voort. "If you walk around ING now, you see all these DevOps teams, doing stand-ups, demoing. They try to get new functionality out there really fast. We held a hackathon for one of our existing components and basically converted it to cloud native within 2.5 days, though of course the tail takes more time before code is fully production ready."

- -{{< case-studies/quote author="Thijs Ebbers, Infrastructure Architect, ING">}} -"The big cloud native promise to our business is the ability to go from idea to production within 48 hours. We are some years away from this, but that's quite feasible to us." -{{< /case-studies/quote >}} - -{{< case-studies/lead >}} -ING has long embraced innovation in banking, launching the internet-based ING Direct in 1997. -{{< /case-studies/lead >}} - -

In that same spirit, the company underwent an agile transformation a few years ago. "Our DevOps teams got empowered to be autonomous," says Infrastructure Architect Thijs Ebbers. "It has benefits; you get all kinds of ideas. But a lot of teams are going to devise the same wheel. Teams started tinkering with Docker, Docker Swarm, Kubernetes, Mesos. Well, it's not really useful for a company to have one hundred wheels, instead of one good wheel."

- -

Looking to standardize the deployment process within the company's strict security guidelines, the team looked at several solutions and found that in the past year, "Kubernetes won the container management framework wars," says Ebbers. "We decided to standardize ING on a Kubernetes framework." Everything is run on premise due to banking regulations, he adds, but "we will be building an internal public cloud. We are trying to get on par with what public clouds are doing. That's one of the reasons we got Kubernetes."

- -

They also embraced Docker to address a major pain point in ING's CI/CD pipeline. Before containerization, "Every development team had to order a VM, and it was quite a heavy delivery model for them," says Infrastructure Architect Onno Van der Voort. "Another use case for containerization is when the application travels through the pipeline, they fire up Docker containers to do test work against the applications and after they've done the work, the containers get killed again."

- -{{< case-studies/quote - image="/images/case-studies/ing/banner3.jpg" - author="Thijs Ebbers, Infrastructure Architect, ING" ->}} -"We decided to standardize ING on a Kubernetes framework." Everything is run on premise due to banking regulations, he adds, but "we will be building an internal public cloud. We are trying to get on par with what public clouds are doing. That's one of the reasons we got Kubernetes." -{{< /case-studies/quote >}} - -

Because of industry regulations, applications are only allowed to go through the pipeline, where compliance is enforced, rather than be deployed directly into a container. "We have to run the complete platform of services we need, many routing from different places," says Van der Voort. "We need this Kubernetes framework for deploying the containers, with all those components, monitoring, logging. It's complex." For that reason, ING has chosen to start on the OpenShift Origin Kubernetes distribution.

- -

Already, "cloud native technologies are helping our speed, from getting an application to test to acceptance to production," says Van der Voort. "If you walk around ING now, you see all these DevOps teams, doing stand-ups, demoing. They try to get new functionality out there really fast. We held a hackathon for one of our existing components and basically converted it to cloud native within 2.5 days, though of course the tail takes more time before code is fully production ready."

- -

The pipeline, which has been built on Mesos Marathon, will be migrated onto Kubernetes. Some legacy applications are also being rewritten as cloud native in order to run on the framework. At least two smaller greenfield projects built on Kubernetes will go into production this year. By the end of 2018, the company plans to have converted a number of APIs used in the banking customer experience to cloud native APIs and host these on the Kubernetes-based platform.

- -{{< case-studies/quote - image="/images/case-studies/ing/banner4.jpg" - author="Onno Van der Voort, Infrastructure Architect, ING" ->}} -"We have to run the complete platform of services we need, many routing from different places. We need this Kubernetes framework for deploying the containers, with all those components, monitoring, logging. It's complex." -{{< /case-studies/quote >}} - -

The team, however, doesn't see the bank's back-end systems going onto the Kubernetes platform. "Our philosophy is it only makes sense to move things to cloud if they are cloud native," says Van der Voort. "If you have traditional architecture, build traditional patterns, it doesn't hold any value to go to the cloud." Adds Cloud Platform Architect Alfonso Fernandez-Barandiaran: "ING has a strategy about where we will go, in order to improve our agility. So it's not about how cool this technology is, it's about finding the right technology and the right approach."

- -

The Kubernetes framework will be hosting some greenfield projects that are high priority for ING: applications the company is developing in response to PSD2, the European Commission directive requiring more innovative online and mobile payments that went into effect at the beginning of 2018. For example, a bank-account management app called Yolt, serving the U.K. market (and soon France and Italy), was built on a Kubernetes platform and has gone into production. ING is also developing blockchain-enabled applications that will live on the Kubernetes platform. "We've been contacted by a lot of development teams that have ideas with what they want to do with containers," says Ebbers.

- -{{< case-studies/quote author="Alfonso Fernandez-Barandiaran, Cloud Platform Architect, ING" >}} -Even with the particular requirements that come in banking, ING has managed to take a lead in technology and innovation. "Every time we have constraints, we look for maybe a better way that we can use this technology." -{{< /case-studies/quote >}} - -

Even with the particular requirements that come in banking, ING has managed to take a lead in technology and innovation. "Every time we have constraints, we look for maybe a better way that we can use this technology," says Fernandez-Barandiaran.

- -

The results, after all, are worth the effort. "The big cloud native promise to our business is the ability to go from idea to production within 48 hours," says Ebbers. "That would require all these projects to be mature. We are some years away from this, but that's quite feasible to us."

diff --git a/content/ko/case-studies/ing/ing_featured_logo.png b/content/ko/case-studies/ing/ing_featured_logo.png deleted file mode 100644 index f6d4489715aa4bc1ecbed9ec6b21531e07a02ffe..0000000000000000000000000000000000000000 GIT binary patch literal 0 HcmV?d00001 literal 10305 zcmcI~bx<5Z7bgi0L4sRgg9lmM-QC@Faad$=3+_$`PH-m>T!Op1yF0-hj(m1^Ro~rT zcQZB9J>Bp3tlz7eu6|*P@)9UW1V~U&P$*K8qRMaW{M+M;@c!-nUMKDNt-*(gX+l)M zW)N2+M-Y^dDcBeUkg_#02PuP$Ox+y@K)g^;(2ka>nh;GnS#A@sEzsyM4baWj{tXQU z#Vg=uZ)9Q(f&h#`=9YGRq-V`-qyS4(K2i-fIYv2q5s-zYq=zF&#Y0}z#KYQz%al}r zAHeIz{l>r+1Tg}*+1l7Sal7%6{(~>~TmSDc11aDiBoJ#p(tj$YDW?b!0Xu>KY(RE; z6DC$h00$S4iJg&^jfoDx%*e#Sz{tkH$WG75#LdXb&CCS&*G2j!&C%41TUk{6U$Wl5 z_(&}v5PNP023J>CpeqXy>}bxw#KpzMz{t$N%uN4ALGR>l2QhM^w{s%PiVgWr7je;4$Z@J$IQ*Krh6V5mH8PPMcvoq7Pa+$Ky8!>V+zd4YR8N_UC%w!DW`Zu5dhG$}8 zV-gi*XXjw&%FYR5WM=~Uw`|Kd+5f__{a>-% zB90&<2-r~-47T}q6ewDNAz&v9usuLTXU;Otk ztl0k%%kZWQ!{0#pUjpafLvPXexBH(oc>DNIB7yAQvd8f)P0GmFWT2qP-#jOz>b7{C z0q=&TwiKL|vJk&=wJL(}+chs#^_`F)IVpOvKR{9`TrvbO$pdo$kW)MW2$4gC6{xdA zSS868YF0*)QeM-e3?Ck}F9c1B*0pDRl$AyLY z^>+)J8XE(cnb1>`@j_)kh*DC*k^aA*`pfZ6R=1$n2!-g$Ms+-Af^2C!zMqK2|^qi#bjR-%5`UM@?aY}Rp^;DV8)5FJ$TqFBymJmA38@Y|v zrCxlDhH24aZ5EXk;-6E+Rc$nX7ME%Cf?KjUHs3KwzB^s+V!ea9S#7vQuwy>&>Uj$8 zP=uBeq|Dp>X1K@TuFB|~HY8~JaCDj=#XzziCdvng*0CgF?}O%(9tuMyjV-F_xJ{4^ z^WiJb{y2PHH0Kk&{C@s~cNCvGBYMh9&JV_yGW1C-!e~Gx4yi)+L36)!?>NA8e;t(? zDq}|iO?ENbgJk`M4Mwf!Mk$$f4Dyt@zuhNjO4CTbppbBPd~|0@63o5T66|I^e20%J zk2eK(L6-+jOfKXgUR}4}3=2KeP%N_6x?hbGAT0%C@BK2dtzT^#RBSp+$`WV5ONc!C z`5IN;VqGd3+ITGcS=43%z=EiOnrjU0rje&35FM0(du)7u!p(XZHR}Dbv3?%Qaq+`T zFT?T-w4^uPO&VE~hUre)s}cI86Sd@;g1-d1u=x*Xa~r7~ER2;lsFV*QkBUePwuNU6 zNxdzPMFzM7rV6Oy&=?=QM#)M8`~qbzCa`x5?vO+F{>bZliRV;0^JXQ`EU}%}BYkul zd9{uQ?~ml0k?x`sxb_w5_`LRPaLx0+OCUv`MsK2%t%#JSAHU^!C_N4d<&FFSMRy!B z4{z^pIqN*}y9}FhFqW1O=c|PKJ{tniuL0N3TOhaQ!#QQ~{ZjAP4O7Hj8wLLc25uF~ ziATo@igms56d>X?3VN7&3Y}K@OOPmy#ISs4bHcv0H!6|WcqNzFwcsXnV!_fNoa#s7 z^e+#4;Yd7FEouF5?B%gJw=yZkf1LbXMQl8}R~iALC&qx67(cf&tk9{{PyWM@)HrI} z&rXdPq2nuUC5URt*!CC-UV9%0 zRTeV<7w36ugn#C)hUWsCuK|;11L+O}x)Hq&fS6!o|J8DpZQ}Rx8jA4=j=?VPXfQho zQp(_p)H;`=YgU^TFPJs;LWS9gxvOV&W#tPK_pug0v0!!lc;1cr#Acajx{f`bI%nLd z-&Q@mfGosJoo3HuCS@AT$ItiWlRI|<=RDVEsOphLP=BBYoAPY&JL$hz+u*-~HC z6CkaTW8%x@<2-~$WnxF!ZWst8k>ozLG8Z@Ts z&s;A>e_HZx$^%G~x;{v>%)sci3R%#4WV>;BudPBG5cke}YH)uzNGh7d;Y)h;$@VT$ zgn=Gvx9QB~)IN2Acjb*dVpWOPxn$*b>*^svWhPJFcgI9p2~}eUEK(n7%sh?AJI%QG z(uP)3`l@obXUsQVr^UQ9FzKG|3|_XVB9MB_UBjuA6x^;yuQaz54~(*G6TY~mqsWBp zI|pN$8z(W$7w@$ z6c2W$C90W}f3io#ePYYPfuEo*{zMMX5K6i6@YzRSBiq;fCD&CBdR9jI(HCI7di8uB z<2h1w?nJAbNva{4m3%R^dT^pM2OJc_ATd)dViAq+Pg%rauSjpVo~r&aLq%?PZtqli z(`@;x24wwVayM zw^m=3|v5ezV8=+{mR?Y3B zS&HP6#vhvt7Er;?*bD(}?5QYwJ)9_gV(eWRTcd#UBMwjqCk6Gi`|fqkAE(`uGYTN z2UIafii63WlB91Lo0aLI0U;M1qRpfPJ)%#oWu?#wmS^R4jEBQ>xoSr-wRPIeY z*bd+0FlQN(8^sS)w95;|YHpO8mW47h8yX>BVsd=#?tiUnjM3TLcIY(Sk#~{~$E%xD z5AfmmJiW(HvI}+ZU*(Y)CEZL)UoAZqSh8920+5I!^!ON+z27bZs zwdpaxMAuAW4`%a0`&Kd`!sQ-@qd$Q53*IMwB-GueWee_ z!7$%osLZPS+rm>+jiLCNJQeg>OLwyr&N1_Jwcs_{)rKuPn0~;;$r)d$?Mc)WGI6~B zgWwXRfw8-{cgU6L*$aew08!>Am9rP^yTnyDh+V8jlza~KQuT*+krsS7e+4eks_AimkW^Hwmq^((}DO3W>{F$?CTfbA&yrC<$e%M>UJ8U0qV*8Z>KHo84=kF zs1)6sX%!jDgr-DSX&d3KBoknI@|c6}$MvMl*(!rZP=|>yoFGv4+wwlTsE__Zjg&XNkc2%pxodkdAP~nzAcrWpc(1QHlLFtGEx-`ktQQ zBjepl^pZJXn96Qc9AywovFLd*E*VeL?z)Bs56`(&uFsvY6tzYY@OiFxRn4Qb zR4ZYMDKCFe{QP_Rl!A4zm_ZD~z(;bS(G1 z#%*4bp6?~+Kl!VbRpW>eg{jbi`+cL9{Z)?2%XSh9L9NgQJ+B88k3BA`Lm}z}!lXr2 zu4BZM!gTJL8zeKjVQ8Kg4~=*t=8r#PRd$0omBY_JIS?ON=S&{#(d!v0$C;zu)K|mi zBnG_sCKVYyVO~C`)7vOX>GCU-bpFpaPgms>_YKRu&r#M6Ex9L7sI|JWw{82W{c6fn zw{Zw93*k7(g+Tx&q^{P^Xa~t~v3%*GLs=f{$eN*ummwXW;4-02kj*Pn?P{=YcPE3r zJ$*iX02F<;i-f9efd50mr<1@R&*)~_#(4)^jVM4~n0MLEOg~Z5b*Y;XljvCh#|+9T ziX|kGZcoKWz9@oEA}9V!&E8($_0JABO*)W+!mZ=mosdXg8?$MX%6K)8x3YLWF}WJ0 z^EXp;b&4IGt6nwq;H%VeAKEtHl^RhoM7{Je&*MVdBA^Afy5jI$Vo^(L4x&9W6f58g z^xh%j%CHH3=h9@xY^D+dD)q~fdlCc%nU5cP)8Y9WwB-l*!1_5*D&})>qY0Jt=p(Tb z^$j@JG~3*vwy=Ck;4q$K?>`Bh+T$46fuw5q?pd8pQS%)f`o8OaZ>zuR>$FJ`R}z4;3B5?!yH=E@67y)+A5g zY;wE0U+_AXQ>c9#lPY(F_7lEnCUZ?J@J-*T4Q7F&raM-M? zlZoX)Vo7kY=qv(^qXka_1sRCe!UqaJdy+sScHd0TVvy*`hs|hX z6akFUbW2Fx*l!fNQU?+zObMP+WZG)FPb9Z&fX^MieQdar*|B0-di%#a72bbhX<5_d zG*uTD{{Z#gOz zN=6MQkD?TOYqUl5C%5D?2u>F`Y^O>#z#OIekyE&7A97&GSNfj9_W9tR=QbuqS_u8e zmlVGj2pN4-M(4D^!t=zzbnO@>qd?|buwX-q(AezIi+n{$aj0o*YF}=C-*((_5T$uT zd#Wd~?IQaCyK1F7kd2S#H>2NP>wUM2;6{NCggq8>w19SR*g+tnHvxSm) z?pJZSbcGXF`ApXTuee6RRt4K3}EMI-L}jqLaAqt8K10na2>S8a|~2mrnJHI zWp{vNpnZdT%^A7k#oXTr;_2Hcy&L&~1SzHhay)}n<5(%+DNl6Z{6+Y7cIQdv(%5z z%tZF?QnHBSTN&|APNY@QHm|gnH>!)!5rPC((;|DA{+#!b4Ji^!d7{~UjUlvpn206} z;=ti;*P-91jxaqK0qS99dsaRXG{yYTXSUP`(M|4nT0T&_gG{k#rhVX;y}+1As$%u37uQMrKaQ+3y|>@>n3D5``Jh@oe? z@%_GN9VG~p(pi*HV0+J4i}y(;x1HR9*{cT0WK4uY!Wcu|OUB7)Rk;AD0&6SKbM4gN zgb%*y)(GkNLLG6B6G}EZnc@fj?H8f5p@x56kKUS$3Qa>VKK&6;H&512za2b(9e~jZ=FCv>sizG9LCX1L+x!r-e*TJsOh>7fOtlMh;hL8r8S{=LeCdGNU7ic$-xZ68zSUH`(uh2mvB_ zTnU?b60+MI5o+PiKR5P8L*u6*>y3$2JmW7JWgp1b`bHTn`IdVeI~u5rC>TU?;C^>8 z-s1B1=NieU#jG3n3Cr@6^6bE}DPXqQk=H`KqnT(PJVGQjuiYuqVA(HxM(!RCxp()7 zu7A*_Nxb%{7NHeUo}60DBMv*;T%k&vu}ajEdk7N-v=?lSeO3sSb0vifa$QFywMd@)sp`GRUGXNzf zQ+(7)R9e3!4R^`3S4B0r6T_Mu2P(0|43rv;O6;j?E`854yb3^AKI!dfBUp(W|5@{R z>9}5(Juca88vd!!uX;23>C#zL+v@XZ957KO9oT+TZgslcR_XLOcDeGLNRw~6lb7P1 zCeFi^IStw^B0533QClpn#kYi(DA#Vb7+)JP7^EL9+N(#~i5Ba`R$h?S6(-PCZNQB3 zJ5s9UTDjp2HO~9REIz?hjj0&}{C06IPsn;fmSSX~4^Gr3!viE8W!~ z9dmTg4Aj#Dr*$oSjT|LLgJp}93YuynQ4{=UC!Dx4q56F4IXov%!#NMul8~=(kek+` z{1$@Gu+#ULgNA>PKN&Q2&FY-tr_<|w+3gA`Q z*g|8rHCnuyy}8M{C9(->?{yzK$$mT&7*OLre~mqfS*(RYck|O2dzddeND0sPT_*NE z6lCSpEtkerp0f>;_HjJdKtK>a`g@*==gl2|`HhSMGuo)JG#ie!b~C@(S?@Y_G6zxz zksD#nzU&_QViTz;)343AME%OVggh6Y%+6$cBWZX7LUIwJ=iF{&U(8TcRGQI0$9a1C zj49=p7G?=tEVMD=Zt0EWLeKCS1aw+l1F@-w%*kaLc9WIzDIjaJWaU#rNbvhtAVT{3 z@uhk|N0oY#V?8^qmcf|K3zh{T$=4o1lbr|H|)M1fGKISEOl+ zkbIVGcUkLEa@e{Alj+Z#FgZ4Che&86!fhRzYcV(LHd+X^@8`%WZ;(~p;ffvchkG}Rf<+WTAtrE;6BiaYcl#Qgb}o#XH|K_*qR1{;++Nc z7N_;7<&wk#m5@Gt2~xHq4=okFPX{y~_d^w6# z%SWj*_2Pqt>@6IpR=@Y)4DXXyC1k>&?$eKda^8L|4iTvOT-WgVTqa!? zxSBMGHHMLKykHjEF*ujveifO*KGfFo+WcHwjZ`WqUW*ej$%=6n;ZiXHI4u0YRXg+1wb6#}c{RRf!-5-z zo<~n7)wriv&z!WgpMv#izazix#&4Hm2qP)C<&(QD}G_vpvJ{&F9fkqzD>Qsv` ziXs#yLDphQsyvG{&tAC0i^s&%=0fu?4S}pvm)1xCfi*Xp3-zpyp;T6)k4|LzLXNP} zQ+tH9=V;k1k!kGRV|~}ytviQ4oSw?-t)~)m7!H5B8*T8eLN$Y%V=NZjj}Q(+-DUjg zc&i#=M12LCo^QJsAwScvy-vE#V|gdG@PDn2p{tF za$c8pBla{qs5k8X-FD((#83M2JAx+SSNuz%$*b(G-&MT}EG&o;AYp0TYqmXXXk6ss zS)3GG&Qz1HINivN{2;gLN2Xymwg|kuB9Eic69$qMDnD-w?V?a)?(Z1>babuVyWixc zx-`F{=ssTZ3Du=issS2bUc0nHJYGhB#~H|KW_fTRO6RZJ6ase&M`ofvRY8v3mcrGb zt>+LNU4B*9s~lXQd?`+4iL2kT>rL1;*f^5keAo0fxuc(_Xcve-`_o*eR{GxRR+#sq zK&d}8-(e|A1cKdo@k8xMB}7RAZnJhwH~aLivBT+o_lgc9p<*s3j&0#l*X&VrNVeP4 z*WETD6r7-(RX!6S^?kO7?@cpAWGkmPfQy(gByvsJB zkz}EPiV=gxpwm>nK)uxE=#)GOry0|^_I%Tnc$xne!h2Q%5BrXHjvI&hpq(KsU*ahV zXq=Pz2D8KT@v1IvY!lbtt8Sd69$QHUw}A%2X25&?nqJY?><-bhgeZS~B;!w)2ZtF~ z%IG@s^2_Qozucov_0k{~Dt!YJ;j5VUy{o!cu+-JX6fAKj7yU!oEY4W$HFMu!%(qh# zU)F~)BKHLl&my}Je5c*vule&z&FcwcKIN|KAu{bL)gjp?jP~DMcF2z!4{O)-eYM3q zj*j_GX2qc@Pdx{dy|S;eVJNdi)I)`JfizGUJ^i4vI1-y79a*)lo|-o{OW98Wdp`PX zDnq1Mxw(jJ^2dKXGMV)4q1S*>)W zN_QA7Hf`d~4TNR@)+0IH{YYN;2U9nyr-`8Mna}|+(+fkZJD1EzaTlk5;IL*_j2$=F&W5QQXyg#vxzFhQ$LZF6@ zX5wA^_n;vRJzc@k>ck*3;&kH>Tv|38=3 z{O@HvZ^GWnc>dP$yp{3%r|QP^Z-vj>ls@ir78Ep;xq&jTo}H22-=aP#F?rDnVMG7_ E0oEjKyZ`_I diff --git a/content/ko/case-studies/naic/index.html b/content/ko/case-studies/naic/index.html deleted file mode 100644 index 89ef6cb8de..0000000000 --- a/content/ko/case-studies/naic/index.html +++ /dev/null @@ -1,87 +0,0 @@ ---- -title: NAIC Case Study -linkTitle: NAIC -case_study_styles: true -cid: caseStudies -logo: naic_featured_logo.png -featured: false - -new_case_study_styles: true -heading_background: /images/case-studies/naic/banner1.jpg -heading_title_logo: /images/naic_logo.png -subheading: > - A Culture and Technology Transition Enabled by Kubernetes -case_study_details: - - Company: NAIC - - Location: Washington, DC - - Industry: Regulatory ---- - -

Challenge

- -

The National Association of Insurance Commissioners (NAIC), the U.S. standard-setting and regulatory support organization, was looking for a way to deliver new services faster to provide more value for members and staff. It also needed greater agility to improve productivity internally.

- -

Solution

- -

Beginning in 2016, they started using Cloud Native Computing Foundation (CNCF) tools such as Prometheus. NAIC began hosting internal systems and development systems on Kubernetes at the beginning of 2018, as part of a broad move toward the public cloud. "Our culture and technology transition is a strategy embraced by our top leaders," says Dan Barker, Chief Enterprise Architect. "It has already proven successful by allowing us to accelerate our value pipeline by more than double while decreasing our costs by more than half. We are also seeing customer satisfaction increase as we add more and more applications to these new technologies."

- -

Impact

- -

Leveraging Kubernetes, "our development teams can create rapid prototypes far faster than they used to," Barker said. Applications running on Kubernetes are more resilient than those running in other environments. The deployment of open source solutions is helping influence company culture, as NAIC becomes a more open and transparent organization.

- -

"We completed a small prototype in two days that would have previously taken at least a month," Barker says. Resiliency is currently measured in how much downtime systems have. "They've basically had none, and the occasional issue is remedied in minutes," he says.

- -{{< case-studies/quote author="Dan Barker, Chief Enterprise Architect, NAIC" >}} -"Our culture and technology transition is a strategy embraced by our top leaders. It has already proven successful by allowing us to accelerate our value pipeline by more than double while decreasing our costs by more than half. We are also seeing customer satisfaction increase as we add more and more applications to these new technologies." -{{< /case-studies/quote >}} - -

NAIC—which was created and overseen by the chief insurance regulators from the 50 states, the District of Columbia and five U.S. territories—provides a means through which state insurance regulators establish standards and best practices, conduct peer reviews, and coordinate their regulatory oversight. Their staff supports these efforts and represents the collective views of regulators in the United States and internationally. NAIC members, together with the organization's central resources, form the national system of state-based insurance regulation in the United States.

- -

The organization has been using the cloud for years, and wanted to find more ways to quickly deliver new services that provide more value for members and staff. They looked to Kubernetes for a solution. Within NAIC, several groups are leveraging Kubernetes, one being the Platform Engineering Team. "The team building out these tools are not only deploying and operating Kubernetes, but they're also using them," Barker says. "In fact, we're using GitLab to deploy Kubernetes with a pipeline using kops. This team was created from developers, operators, and quality engineers from across the company, so their jobs have changed quite a bit."

- -

In addition, NAIC is onboarding teams to the new platform, and those teams have seen a lot of change in how they work and what they can do. "They now have more power in creating their own infrastructure and deploying their own applications," Barker says. They also use pipelines to facilitate their currently manual processes. NAIC has consumers who are using GitLab heavily, and they're starting to use Kubernetes to deploy simple applications that help their internal processes.

- -{{< case-studies/quote - image="/images/case-studies/naic/banner3.jpg" - author="Dan Barker, Chief Enterprise Architect, NAIC" ->}} -"In our experience, vendor lock-in and tooling that is highly specific results in less resilient technology with fewer minds working to solve problems and grow the community." -{{< /case-studies/quote >}} - -

"We needed greater agility to enable our own productivity internally," he says. "We decided it was right for us to move everything to the public cloud [Amazon Web Services] to help with that process and be able to access many of the native tools that allows us to move faster by not needing to build everything." -The NAIC also wanted to be cloud-agnostic, "and Kubernetes helps with this for our compute layer," Barker says. "Compute is pretty standard across the clouds, and now we can take advantage of any of them while getting all of the other features Kubernetes offers."

- -

The NAIC currently hosts internal systems and development systems on Kubernetes, and has already seen how impactful it can be. "Our development teams can create rapid prototypes in minutes instead of weeks," Barker says. "This recently happened with an internal tool that had no measurable wait time on the infrastructure. It was solely development bound. There is now a central shared resource that lives in AWS, which means it can grow as needed."

- -

The native integrations into Kubernetes at NAIC has made it easy to write code and have it running in minutes instead of weeks. Applications running on Kubernetes have also proven to be more resilient than those running in other environments. "We even have teams using this to create more internal tools to help with communication or automating some of their current tasks," Barker says.

- -

"We knew that Kubernetes had become the de facto standard for container orchestration," he says. "Two major factors for selecting this were the three major cloud vendors hosting their own versions and having it hosted in a neutral party as fully open source."

- -

As for other CNCF projects, NAIC is using Prometheus on a small scale and hopes to continue using it moving forward because of the seamless integration with Kubernetes. The Association also is considering gRPC as its internal communications standard, Envoy in conjunction with Istio for service mesh, OpenTracing and Jaeger for tracing aggregation, and Fluentd with its Elasticsearch cluster.

- -{{< case-studies/quote - image="/images/case-studies/naic/banner4.jpg" - author="Dan Barker, Chief Enterprise Architect, NAIC" ->}} -"We knew that Kubernetes had become the de facto standard for container orchestration. Two major factors for selecting this were the three major cloud vendors hosting their own versions and having it hosted in a neutral party as fully open source." -{{< /case-studies/quote >}} - -

The open governance and broad industry participation in CNCF provided a comfort level with the technology, Barker says. "We also see it as helping to influence our own company culture," he says. "We're moving to be a more open and transparent company, and we are encouraging our staff to get involved with the different working groups and codebases. We recently became CNCF members to help further our commitment to community contribution and transparency."

- -

Factors such as vendor-neutrality and cross-industry investment were important in the selection. "In our experience, vendor lock-in and tooling that is highly specific results in less resilient technology with fewer minds working to solve problems and grow the community," Barker says.

- -

NAIC is a largely Oracle shop, Barker says, and has been running mostly Java on JBoss. "However, we have years of history with other applications," he says. "Some of these have been migrated by completely rewriting the application, while others are just being modified slightly to fit into this new paradigm."

- -

Running on AWS cloud, the Association has not specifically taken a microservices approach. "We are moving to microservices where practical, but we haven't found that it's a necessity to operate them within Kubernetes," Barker says.

- -

All of its databases are currently running within public cloud services, but they have explored eventually running those in Kubernetes, as it makes sense. "We're doing this to get more reuse from common components and to limit our failure domains to something more manageable and observable," Barker says.

- -{{< case-studies/quote author="Dan Barker, Chief Enterprise Architect, NAIC" >}} -"We have been able to move much faster at lower cost than we were able to in the past," Barker says. "We were able to complete one of our projects in a year, when the previous version took over two years. And the new project cost $500,000 while the original required $3 million, and with fewer defects. We are also able to push out new features much faster." -{{< /case-studies/quote >}} - -

NAIC has seen a significant business impact from its efforts. "We have been able to move much faster at lower cost than we were able to in the past," Barker says. "We were able to complete one of our projects in a year, when the previous version took over two years. And the new project cost $500,000 while the original required $3 million, and with fewer defects. We are also able to push out new features much faster."

- -

He says the organization is moving toward continuous deployment "because the business case makes sense. The research is becoming very hard to argue with. We want to reduce our batch sizes and optimize on delivering value to customers and not feature count. This is requiring a larger cultural shift than just a technology shift."

- -

NAIC is "becoming more open and transparent, as well as more resilient to failure," Barker says. "Even our customers are wanting more and more of this and trying to figure out how they can work with us to accomplish our mutual goals faster. Members of the insurance industry have reached out so that we can better learn together and grow as an industry."

diff --git a/content/ko/case-studies/naic/naic_featured_logo.png b/content/ko/case-studies/naic/naic_featured_logo.png deleted file mode 100644 index f2497114bf40a1a0fad1eaf65fb52b2c03918f08..0000000000000000000000000000000000000000 GIT binary patch literal 0 HcmV?d00001 literal 9433 zcmbt)byOVBvoG!hTb$q?g1fUg1Y0!t;;xJP;_eax1cwa?kPw19A&}sb0KpxCTX5dw zdw;)s?>q0l_s`pNW@n~ORZn$Sb$8XLI$m2-86SrV2MGxYUsXi`0^ENBR~Qxw@PDS= z4FztLFhwJnuDcz~59(!$BxmDpWlOK>3bnU|*g|arJcn#0k&sZ?9rcW0Mj9`~t=(O@ zp?`U}{arl(X(S{`8GjF`wX-dZ-pbbA(M^i+sI7~U-qA*i(NIW(SHnZz*1=IF(92dg zP*cx3(AiqdhEYbEUeaG2Xu#DL2Br6Rb#e0+_m^V)r(JR2`7fD=k^UbRn6nh)e<)?7 zp-nIE?qy3a#4Y^Xnop3IUPO$WPncIwh>we&pO;UBhgXP)SNJ(EpE$3mI6oi#zkZCs z*Su`(#32ew|N0j2CdKFggL#Pa@c8-par+5yyL;L5@QI0u@$mBV@bf)5B1>T z6GyZd-q1NN|E`nNDJcK`1W{bdBS5m#`x z_Hng!gQ+S=F#;5B8%G;)UO}*uyn?&{m`|9Wk55rlOjtlpQACa(tSBZA782$Cx2&q0 zHw@}#ZToLo$N!S$|9{Gg%X`^EVeVdf?(Q!Cww|_wJIvkN!QF#iUj83z(X(qntsUL| z5;^`3@W1r6^>XyJwNdhNccuSlgv1^H2QBz{6~#mqg!#qfgarTT0l%n#qJWURl9G^s zAXos*_%B(T|Ec$iyov%MO1xrnA_D)TEDxXz&tLQYH*^1+1o+V3$NzL*;N?Hv-_{NA zc`v}ZNxDjbS;nBOsvxK5KfiB==?_x8ebbR|9u86g)60Uf(9*_VpKF(MV8w+CIzpu7 zu;bt=M4-C|Ms|v~5Of_-JrYhZ2%QK7#$_R#3~KwHegAmLT43+hv|z>^vD2(Phw-lX`SnG>wZeNg&P8iw2KKJm2jN-w^u@(RN<0e6AzcpJPzENlLf-$b zPzvMHh87yXQ_%*cnny|i*Xx5@zs#aEt0GV)v5G1!N}`BqsgWKMYk`NVm1yQjTcEe= zZ5(5K=SQou7v{5&uSg( z`E6h_X+(qTFJS@cRP)E1-*rFVy%gZ$8h%&#De|+++}i5uD!bCf@1J!B z?~Us^Z=B4{i-a~ygaC6S9SXY4YF{1fIYr_OZ#W04{O~p0Pc*=Ll5R(OX87^kuOZQ_ z-6<+2X0atC=;T*ojEs!Uc)0zsi2rzaL`dKh%5lF{YQA_5$4-XTAx_ocltXiAi}H0* zt*$nUcf|BN#>rPRY|&zy8>sewXdh)$v)fVSNS~X0eo-#_TG(~8NXn$BDERUIk5u3} z??aSq6VL~4?$LF667i_v@rPBzVgYQwrBl~<9N1SwQPoJ>sXb-MTx znJ3h$ZV9tNHt{{^dg(Qn%0)X+&4Qrf`sy|8c6+# zP6LONR2!DtU!_%Zb7l?MAR{ANDKgq^%Q82==dO2Q z5&k(^PEBX8LJ?ME2-f`yCR}*DI~#izC!?yWn#rccjqAyT9DF(JCd+GajFJn}X`eL0 z_;Wb+LTZgbR+}?BCMM?LxR0RjMPWvM{^`%>JtHa|!sP~{I<3Uc68s=&q9v7&q>DR96KV32ypPzK) z*72~j50Ho!6FeI(e1Yuw5Tf=i!HjtFT!!(S|NFmr8Hrk6W2A3Pb-WE2}> zK9de0RBmRR^xfIN%BK`V?{w3A`SK;l&=7{%jyetBnfBa~=H%3rV(sBv!`H+7#f1fj zZ_hs!TYJ;#xFVXH83*U)<}$S4v@BCPO`&>c{uQ`Qb*UdU@DII&wP;6`oCWja;^O3E zI}3%@8PH0u?LstKUx1Kuy)t(iqi%1X5ov-~_`O~R?B0yI@51jE+y}naIEY5Xylc+r#UVfaMoT;&v2g#CC}ssKK~MdvFFue_zg)Wc zQTx1x%F6k9 z5NLU|ODMiT$C#Ly5M}ps_HW`rzna|NJk1tynvSCt6&76T2=d$HRkq0=h!Nw8l$foI zfVo-kZgO=X9A|c6SP6cIv!SOO@fo_PWNEA&Dr>fkAGlTYJ;W4kHql4oT_QP8YFYXR z^Gi8ndNm8hS08t$s=lVu@j$qknKugkmOeC|vTEe7ggH4mK_D4rRKQYi$HT)z1De07 z9rksa`&cn|=IH3y_4v@%*jVlVN}{PeTCyq+wEIxGU>SXScD8JC?Kb3&#yHWS0x7_c z=R-=&YWEer2>)1itb3@?9uC$gm52WN?24eSHK^GhdCew$ZFm~kEq;1(;<(f<#@9aY zHSJ7%MIgv+-e5PfH&gR0OJJtPFxJB1auV2u>>`zoWLbr>8_%*AhR7Fsy54;%XRxCnt7YZ&5o90w#u8%!od;0Sx-GVIq7vLpqt!f$q4*D+RwW>lBk*YE{Ef3BTD0` z_H2u{{aqJqH{)bDnObDYe;q&FWGVP;1X!qk7lOy@x*m)lDlhkD5^HTrHvZt%GJke-0Z%^S1!!l=T(+#gVP%w?8YFdi&?dU5}0Q8;!c= zaiN5t8Yqv{Q?)8TgLiB)fpBAOFF`HzVK=X5Mmw#tBl-^`F-L$^b%yrN;i2Ys&G-4- z?V=#f4W%8hGX!;)JppxIZ627_O-&+CDm4pln$U7^%2rlZZZ3Ym?m)A3=|DPv%i@#q zDxfTe_i>=n867T#r4Z#im89PmHr_IJ60dTrIZwshfe5K6-7cr3VpxwJhjo zP{~AcTkbR)_4Peky06GJ@9k;v-u_W*X7WlG({8sn8rLA-x}vJQB9*yf@WtToy|uv0 z{l0~iRvj>Q(H*spLqSFdFiqxa>N+ls)e?++)2^mH_4Ghegj!XNtlYwyRRq++?=nJ2 z8_v(hWK1Y?r~hog7n>!MDT5ihc|H(B#1KF8H!6PdN}Oqdp>4hg)izZX6|bIKo_7yJ zY7E}njTO1>IxV*P?8=EXvMzPB48%h7e~fD*9Z;>JL7@f5cBC)wm9E>w@0uy*)Lu znO#d~Ql88^@A$J+^9dQ)B6GkbD)7d+eMxMOO%Md~BNpc4(-HP)j%{z5QYiH_16}t* z8|z4viwh4k?BTjfFSd;f!#QNpYvQ{c40j=pP zR6=56D~`Uh>0me(8aCJaHcCaTjAtM*F#>Y-o0v4V5tes?ATSXp2#4w-t?Gi=7BzmZ~i}IzhCq zNEX0W>6&uRivwAM7h2#BnOk=dWW~0+RZ8`F_LfLzEJ@g zG%E^g${R^s*wTwSab_Fcl=GEWh?)8BgCfI+L@A4$izRW<2;^$sr_JG~o8mif*Whna zuuq>Ve0$aOoST~)LvT;MO6)F3n}Uc)j=*KUTlT#{;#eN+aN!kIV5{3m3fH5S%suEt z=m{YKItBYU&BR#_Yy3CWoUIyH+|NkQ@!wQzkrJPIW>@7xcXK+8ur!+4h~h}y(jB)< zt*lBLDb-kgDLwnbkx{uwjk-ebWALeWKD-}#4`eaHCtt`B@{S{5>o$AF16&P$0Ra+X zViyl=p$(-cb!qwT0Um{|59WtMa{=YegYzt`kGy6o78&^=A$KZN{5C48ta0expPgnn z1G%K>%#BZN_}6>wpKVuXt5x)gt^fY*Y}#R*F#go!PfJ_jnDd35seIha!j%=3o43Z< z>Kh4RPjOxgI-;19cq0UVq$!nAinuK>B_gk_@tCPi%r`pk-dc&~E{FVCT(=5oup{fr zI-k;#nKQ`EVg?;Y8w=Ey9RG+~pZDY6@Y*gXZfPj2)7lxZpz7Q!Nn_;;-(O<${#k35 zdOqcvUl$Y03++XF%UrD!YRbF@njhw7@e@)H6G;67$%$e3>gkAZTnxG-R=SzQhWpP+ z(<8%;*S3cB6{2)MunV%fOP?+lVqF|9TP#SB}t@W!pHO$Z&PXld!q? z93?hfGsTO`*b6wX0Y22%M~%O`l+qs}kHa7RQeJ{xSElpBzMe%3q>`DtGlp3)WJZ=@ zM4)~PXnAvME5Uu%>0WT;^5EC5csCF;y==)@;c45__VV%qN7-CX0=duofFV7yxVFPR z{+a%{7gs&Nk~eDhK}mLb^?83&tt%4z^!vC-4co7A{m%k(2&{~tpT|xUp3WGYgweIN zwdeX$!3SQIqhALG5}MTvIxmiX*3lJW$A0v*B(4cTN{);3@b+$q0td~n_AUk#%lKQ6 zuZ=ohPb!JAu&^|T&YreR0bwFD+t@tC7hV$8$?G`Fx#(@sGK)%%IX)}f&kQYd=B6={ zqj1jfUh|H?35`!Q8GJyfsoj9PN|XNw~I+AXE3SrI5Z@x`q(D zN$3aHK2~U-P8oXz3Lwl)X7}sq8*qx#Sl<%%4PbB9;VXT6XZG2vrPlNH^x5{R_a+j!vM!I3-KR9n`cqyuJ~qZ9E}l=@gAbD^6n!>}phXtH(Ta~8AGfj(Z&fvYzw&DkPam%)%pT;%#>&>v4 zer`*(dP%KXJ6W}Ty>gXYGh|qYMT-9t>?Fjv>-RR?!hdI+p$yU6^OY`aKh5cTWsfw3 zoZGB*T2SQ2&C^wliyRfrb=MyVEk%sS#7$D)Sd2^~b&_j=bvH6gug#2r=zSq(rtwQ>WN=k6-)T9v!*iJpL zK}q~MK-(}C>s3__)onhkc|)zAd4q}8Lq;vnj{TLW<1jF%QnoJ{E8ETA_(NUnS`FpXr8hWNkG6ktqDOH7CpyUU>GbYlDJ zBwj7SBl$vqg9kTLn;c6BA080OacGtCn6)aLHR*Gb7K}osLU}tA$$19C*0+T0*2e|x zM_=Ue_CSb*@ceEs4*)2DOYvg`*6Nd1-#>?D#qHBd?zFVD0H&erTJf=;PSpeIzR>)^ zV-=jZnp?r6A=|XGsB-tEXRdAQ1;vl%?X>Ch^H0-y?+u*urLZk~L-f?*P2s4Tb=!#- z?UP>?n+%E#YOY<*Dix2y@^bHvD-5TWZ1ypr7U}JUUJ6TX|iqr9~e}meI=vJ@P?Zi$$=ZM5|R0mt72+MR!}o4kL9P_u^s& zrqD_m3vc7>Oi&LV+R4_f`%3t#@S8k$lGndUP~jnc+TpcRj*#AaWDsNR$%`k zT+3)+J2@sH;|jVVs>_E9VLoKww>GSiSX1IW3K>Q9<-buq<&VNL>@MIzEWS`VU&CA8 zjM4b{HrlmTNRj!%Zb&`*n=x&-_7zW5nfYv<@pIz$hL`goY-)0))ItDe*0%+Hbfi+M z%qsgFh8lQ#-@CK3)7uM9S)WCc6n?vyg~q8G(T19NbifoP;)Zi8lz-pw1Av{o{x`fi zkYC&X;Vz)nj*pvLb7P&i*=EXWW8}5S>(D-vvWWtzpdTrlzk!XLQKL_1=1c zsMu|0&Kv@R!Bj#{(+FL*-#-hd-|DL)^r+6RA_hIz2V8-)dS|>uSjc_4T8A(dLTBGk zQG6=yvlkT`>$f}krLs<97bYHFEEkCZzxsrzaQw@R}R9LYjFG+Ci;Kay{2 z!2+a@ZWEEzeuF zi`v2-A1XCo?__ANWQTg@QTQqa9j`=u7$fYL3c4K6+Uo9>wfw0U96uBb3FkD%rfu>& zGH-n+DlNSPn_}8&0iuNaZJEbw086R)`CF?%RL+o*oQzD^eTl}Q@;#M+9o~#;#hW6 zCUBhHxMGVPqD8dd0C-VdZtB?X6co?xl9-%q))9CCM109x;G$yAft07}lpFi|?cUp( zmYx%(ip^9YP`@XQ9_TPRA)!>AVOMJbC3w2dg03uD4G26egTDdWh>ksAW6!$@WxDID zD?D0Bih)nS?gNEFdH-0l+sJZe19{F+(B*0@-5Ny)5|_a{BV_in6EI_^F}~nB zQm!SY-DrVoS21$%3i0A5_4e+R4#5WUu)@f4*rODu9JQ&bX;HuyrIk@_G(Uiip)Oa? z%Gw2ogm@B*hW>uv6iy*o;lA8iO_-#}H*1A;07U%S81~)60A{{=6-{oA6n0GSfguA; zZ*OluZU%O$QYL>!PGz2M!K~Quxn!1Go!Kfax_WHXA9J$j8vu|M2knXiB2rWw(Z~nu z_FNtdb0F6S$~~r10MvX#MnclyzARlvq5TvoJEkp(y}!9PU;eQte2oo**76Io*(|cn z3Pl-~nR@mnfy|w(lzIVXZjZAu!3c5;8c2H7`M@i$=)Eg$r? zn>?_?9}a#s$%{chC9y}ntRkcA$dmBnHi8B+b;t6&qk~ zdTD`dL_q-!M?3Kvh;~+Fro~3bg{D8*wf@BM$0~=5Ws!E1NaDkiKc>5RM=EofHm6B0 zED)6@3InJ!<|DPROWa#+>=_$T3~Nk$s;EvR6YWVoCII@uy@__CkLg5(TlNA_1i_oA zL$2^ArR#(gKG%bxiFky?i)pCr!B~WWDN6D|-#5M3J+KHB5-1Fcdw6@`NC=oBf#{Pt z#U_S2oM@DrBVZC>v_-I8nOCYOcS|ocyUg)>@G>wUbH$wY?SL)d5<4lum83x7>uOr~FMsZKlLaB5?4*aH54VyP?l=KJ4P?x>`ZY0ckZ9%!M(_tq zX?IXyltp;73DglX=R{k$AJMvc^Ich=_4AZJOWL8m1_}emoYgg$m1TYPqObsJe|+w; zFVmh8_L1fh1Ad3St~a@&eWe6OF$Rnqolh^E(h{tVoollk;i_P#2t>kgR&>br}tzU3@f?qvH3QX z@>HT!*eD6|jNln|^vXH_K=0BD9&6ImQG`dJViNbo9(f?OL^wr7$t5#Xqa~Jb(wj7+ zWn=sZzqQ0!Y(wRY{YWZXLWYb69R09NMN*(Aw(~sQ7_wxHp@|TC0FX93X{2Fx8**d( ze2gbJXrNJ?W`dYfeOUtkSf>bLJWj!$i85vUW%hQNpi7rH{L*LaNY!W#SROPY5RdUj zZy4NW+z~yrh1Z3Yqs;_J+*iuGHXmWLc7943fZFiS>6*YX353uXNC}KaBQS7Ej$Wm z0==F*70D3MU0hgi=+Dh1bZ4pZv>ttoz#p8+tAr&n9{P!d31u*aMvQn2ZeDB@RT_OH z>ba({fa`5LGbZ^%%&HaMzf46zFg&&-I-8jw=X~@VR*J;ceK(?E8RP>aSYGcoBjrEnSoFMWt0{0 zNOkZ7^4ey!=58-!ZVKW8v`!s?`V_ljeQmsIOlgumH0g1SPn6A3LJBp-TA6(=lIH$<9G4aIO+^LOmGfA=-Tuy+0hI7|~2b69Ywt zS>lYVoZ~i?F=?wiuoXo+LCYq1C5gTc_^uJtKbe4|BK< - The New York Times: From Print to the Web to Cloud Native -case_study_details: - - Company: New York Times - - Location: New York, N.Y. - - Industry: News Media ---- - -

Challenge

- -

When the company decided a few years ago to move out of its data centers, its first deployments on the public cloud were smaller, less critical applications managed on virtual machines. "We started building more and more tools, and at some point we realized that we were doing a disservice by treating Amazon as another data center," says Deep Kapadia, Executive Director, Engineering at The New York Times. Kapadia was tapped to lead a Delivery Engineering Team that would "design for the abstractions that cloud providers offer us."

- -

Solution

- -

The team decided to use Google Cloud Platform and its Kubernetes-as-a-service offering, GKE.

- -

Impact

- -

Speed of delivery increased. Some of the legacy VM-based deployments took 45 minutes; with Kubernetes, that time was "just a few seconds to a couple of minutes," says Engineering Manager Brian Balser. Adds Li: "Teams that used to deploy on weekly schedules or had to coordinate schedules with the infrastructure team now deploy their updates independently, and can do it daily when necessary." Adopting Cloud Native Computing Foundation technologies allows for a more unified approach to deployment across the engineering staff, and portability for the company.

- -{{< case-studies/quote author="Deep Kapadia, Executive Director, Engineering at The New York Times" >}} -{{< youtube DqS_IPw-c6o youtube-quote-sm >}} -{{< youtube Tm4VfJtOHt8 youtube-quote-sm >}} -
-"I think once you get over the initial hump, things get a lot easier and actually a lot faster." -{{< /case-studies/quote >}} - -

Founded in 1851 and known as the newspaper of record, The New York Times is a digital pioneer: Its first website launched in 1996, before Google even existed. After the company decided a few years ago to move out of its private data centers—including one located in the pricy real estate of Manhattan. It recently took another step into the future by going cloud native.

- -

At first, the infrastructure team "managed the virtual machines in the Amazon cloud, and they deployed more critical applications in our data centers and the less critical ones on AWS as an experiment," says Deep Kapadia, Executive Director, Engineering at The New York Times. "We started building more and more tools, and at some point we realized that we were doing a disservice by treating Amazon as another data center."

- -

To get the most out of the cloud, Kapadia was tapped to lead a new Delivery Engineering Team that would "design for the abstractions that cloud providers offer us." In mid-2016, they began looking at the Google Cloud Platform and its Kubernetes-as-a-service offering, GKE.

- -

At the time, says team member Tony Li, a Site Reliability Engineer, "We had some internal tooling that attempted to do what Kubernetes does for containers, but for VMs. We asked why are we building and maintaining these tools ourselves?"

- -

In early 2017, the first production application—the nytimes.com mobile homepage—began running on Kubernetes, serving just 1% of the traffic. Today, almost 100% of the nytimes.com site's end-user facing applications run on GCP, with the majority on Kubernetes.

- -{{< case-studies/quote image="/images/case-studies/newyorktimes/banner3.jpg" >}} -"We had some internal tooling that attempted to do what Kubernetes does for containers, but for VMs. We asked why are we building and maintaining these tools ourselves?" -{{< /case-studies/quote >}} - -

The team found that the speed of delivery was immediately impacted. "Deploying Docker images versus spinning up VMs was quite a lot faster," says Engineering Manager Brian Balser. Some of the legacy VM-based deployments took 45 minutes; with Kubernetes, that time was "just a few seconds to a couple of minutes."

- -

The plan is to get as much as possible, not just the website, running on Kubernetes, and beyond that, moving toward serverless deployments. For instance, The New York Times crossword app was built on Google App Engine, which has been the main platform for the company's experimentation with serverless. "The hardest part was getting the engineers over the hurdle of how little they had to do," Chief Technology Officer Nick Rockwell recently told The CTO Advisor. "Our experience has been very, very good. We have invested a lot of work into deploying apps on container services, and I'm really excited about experimenting with deploying those on App Engine Flex and AWS Fargate and seeing how that feels, because that's a great migration path."

- -

There are some exceptions to the move to cloud native, of course. "We have the print publishing business as well," says Kapadia. "A lot of that is definitely not going down the cloud-native path because they're using vendor software and even special machinery that prints the physical paper. But even those teams are looking at things like App Engine and Kubernetes if they can."

- -

Kapadia acknowledges that there was a steep learning curve for some engineers, but "I think once you get over the initial hump, things get a lot easier and actually a lot faster."

- -{{< case-studies/quote image="/images/case-studies/newyorktimes/banner4.jpg" >}} -"Right now, every team is running a small Kubernetes cluster, but it would be nice if we could all live in a larger ecosystem," says Kapadia. "Then we can harness the power of things like service mesh proxies that can actually do a lot of instrumentation between microservices, or service-to-service orchestration. Those are the new things that we want to experiment with as we go forward." -{{< /case-studies/quote >}} - -

At The New York Times, they did. As teams started sharing their own best practices with each other, "We're no longer the bottleneck for figuring out certain things," Kapadia says. "Most of the infrastructure and systems were managed by a centralized function. We've sort of blown that up, partly because Google and Amazon have tools that allow us to do that. We provide teams with complete ownership of their Google Cloud Platform projects, and give them a set of sensible defaults or standards. We let them know, 'If this works for you as is, great! If not, come talk to us and we'll figure out how to make it work for you.'"

- -

As a result, "It's really allowed teams to move at a much more rapid pace than they were able to in the past," says Kapadia. Adds Li: "The use of GKE means each team can get their own compute cluster, reducing the number of individual instances they have to care about since developers can treat the cluster as a whole. Because the ticket-based workflow was removed from requesting resources and connections, developers can just call an API to get what they want. Teams that used to deploy on weekly schedules or had to coordinate schedules with the infrastructure team now deploy their updates independently, and can do it daily when necessary."

- -

Another benefit to adopting Kubernetes: allowing for a more unified approach to deployment across the engineering staff. "Before, many teams were building their own tools for deployment," says Balser. With Kubernetes—as well as the other CNCF projects The New York Times uses, including Fluentd to collect logs for all of its AWS servers, gRPC for its Publishing Pipeline, Prometheus, and Envoy—"we can benefit from the advances that each of these technologies make, instead of trying to catch up."

- -{{< case-studies/quote >}} -Li calls the Cloud Native Computing Foundation's projects "a northern star that we can all look at and follow." -{{< /case-studies/quote >}} - -

These open-source technologies have given the company more portability. "CNCF has enabled us to follow an industry standard," says Kapadia. "It allows us to think about whether we want to move away from our current service providers. Most of our applications are connected to Fluentd. If we wish to switch our logging provider from provider A to provider B we can do that. We're running Kubernetes in GCP today, but if we want to run it in Amazon or Azure, we could potentially look into that as well."

- -

Li calls the Cloud Native Computing Foundation's projects "a northern star that we can all look at and follow." Led by that star, the team is looking ahead to a year of onboarding the remaining half of the 40 or so product engineering teams to extract even more value out of the technology. "Right now, every team is running a small Kubernetes cluster, but it would be nice if we could all live in a larger ecosystem," says Kapadia. "Then we can harness the power of things like service mesh proxies that can actually do a lot of instrumentation between microservices, or service-to-service orchestration. Those are the new things that we want to experiment with as we go forward."

diff --git a/content/ko/case-studies/newyorktimes/newyorktimes_featured.png b/content/ko/case-studies/newyorktimes/newyorktimes_featured.png deleted file mode 100644 index fad0927883a9310b252dc3ea75172b7bba5a9e0f..0000000000000000000000000000000000000000 GIT binary patch literal 0 HcmV?d00001 literal 20189 zcmeI4cT`hZ*YM+vXauDwB8ngcML?2}(EHGf^xh$a009yp3B8R7BT{4p0ckq)Cej2% z1XKnP5mBm$fT(l<5ow`)7j-n_I8UDUUGMi_vQjqZ?sCrW?0xRJcdg4seO>hf^vCEy zAkYC#4HZM+w;%X~f`0m@RuCKg+%g{hQob*ePO;LFb{$=U@j#E7e>O7NGPBICHmt@C_gBk$otvJ zUw%|DL^J{GNy2*Ixp(}c96Y>9auCQ)ps%kl-GiH^?K*gxzq4)?>* zM3SlxfZz+#A6gQP{XH>oLk!Wwn}Eis`T#@W{n6-2Sf_83^EXdBNB*_%7(eX4nC%?- zZ1yP?KF^RWFji@00tQ9$AQ*dixXXXeGX1|0aw{wEWR)09jGJEvg~sA{$O_29zc>3f z=ifYHR8S<0{7z0ng{6SEF+yA#DJd?`KFjtdsu_e25~Xk)o77 zC^WzK0o0HqiiG;Fyin>=_;Ex_J9!XrD3UxDhjPZiJ@L-6@NbP2nqMj zqLNPHjz|dv%JCCK9UV&Zf4ElhKzr|0%oo>=|3lXl=KpZ5N5BHB9m<_TXQ%i7(-1#! z^N&y(SR#<4{uH?jWbmi;%n0N4_tw7-xMM%9OP(kK5wlZua*)5u?(0JKb)DL=`Bc5q zDD=*XCXe1J3Jd`FZ>`_D_0{r=8TNlWfuH?TnE!hWUl$DiKNZwxSD&>0-3igdiR6nS zU=*E!ocM3+{j=x4n|>cc7QS;9aL4|AGCsHdeUARh+JDZe?9SpQ4desRH`J$jK*+-X z(e^F!XMOXXd3E>xq%VOK23~N)&i5Bn%ERBATKs2I%ERBAezP>eV@dMJFVVhhrm)&s zfj=(|M!;qPBQGu{CL$&a|GxQq3w^8~#@$>63)C=iCo_?Vzl^>=@sX)U;NrS3P!$lPlr2-8XB@L=F4Hs2JlnOLl zlr*TyG+a~>Q7X`IQPQ9)({NEmM5#c-MM;CIOv6PL5v2kR7bOj4XQE?7ga=*3N&1lG^olnTvQQJ zD$sCI(x58Sa8X4>sX)U;NrS5VA92xtdk7GN2cG=%1s?I+E36|3JRZo6)-coofdWo} zK);>`f!4Nx-zgBtTNnhIwg-WvuYy2FJmT$|)j%LN0ZkP}W4{;g(w+TSd&&1YQ&MIJ zhJM!(*#|v*P>Ej&4AoHQriXwHcjuVv@4^b|JX%@^V417phD0XYswfp??mKz%BrDH> zQ*CNS;%CJ%>+h~CZVRJgACB4j$JowoZw9wJEkBkbhAduNxht~I$%1Xyz3cRJ$-sk- zyFfy_?zvjSZJnHLw>~X9tG_Z+13GI%7q{fK^~x zVuUG-`}Pn%R&DWDxp$$Vh=>XQ)Vvcf^%R*OwYO*J=9iY1YUO@dT1u3!8F0s_p^c4i zDHZzVKy#8|Hsm@9EpY9^knf>Zz2{{iP5pZr-;R&>SqjnTc1?LcJRB7qxH30fj#km4 zZ+Y%inIRi;*JP#-net8|;{m?LV{2nkPR6&Jm6PS@Xof1Ws&*s)`dfN`Ma6TLxdvOh z-A3l-=7!epH*baromA-x%PJt>xfAtBOF^OWLZg9Bo?M;JUb?*!^{3|*S)1YFxBzU+ z;}Sb}sWOkykP!J!FY&Zklp}j~RdIiT*!26zNcB5@7cG)Dhd9L2I)0A7UQoc_@9G^M z8J{f5>ox-86A}`VkT5IMw|;!b*vxF@NqmjR%hkD_+Dg|dvV=tgX7Nb$5g~g=LatKC zhsRR|lP=-E_quY5xRNxzG#w1HO74@JtKmQY^ixRk=-sX9bdmfg>>CH_mS;6JG+s&G zeC%RPUY3cVqni-)s5x%t%T&UE?%2R6+v*whm+&tLHnYSS~> zbzdP`z?a#emVf$0W;1(}jmd$b!J*S1r^2{5KE8=lR8%Y)A)MMyUVLDsstUFG4Kp^L zF7N9zL|*!N&wiKg>R9U*RyMYr+}xCo@gvS1IY-AQFY^Za%+v3u-zZsht{ezhnaT)R z)^=36-P+U2|o4t1B0TwYR_5=E;-g#l=8Ral*7C`AQ+&(aD)pHvW!F z+v~ZDL+8(jrRz?t5l61c`1ZA`51y7f%5#O>k-62j$(Yae+ugFuH{%oQ-2Ok~A#}!JJg$&$E$*xXrNUERo!rfWGC%Ljo#goLyU z4g320&$d}#mk$;XURcZ7*U1pZ9r;?>I``?*r@!9&#ekJ7==l|_k^d!z0}S0$OA8AN zr48cwd3k*6UAgh`@ljDx&aIcCV`5H*_gD(~)Y)u;_dFP{&@VAnS65GFO}>(NG$CMF z2nO3)epidqV+_PRIj+c@gQIs1VvJT*^qV|%fL&C0Y%1T^eWW~KtS&eqp~|x4{>blF zdz4H2M^;vWvE7yQeld2&EM$>OiP_KK9=pi?H$yAC>F6}qi!QXcX(R-`*>!=Aj?U}& z%F@;*V)jMrt>{{xrD@K^3`OPN_*v<%P2_Ib$K-3Lk$VdA4{U99OlU36E~ge78kw4k zFQ*{9IHj)^73Jxs+;JUyghUT)PJ~6$*FJDWZW0I09+bx1vcO>r3w_dN<&%ex-P-WK^d?pOHgWw@Omqrj)}c!;{kX_=k-K08ylN$Cb=8V z^VjMJtMToCTa2Y?rAuh?uSTopB;SP+&05$q}s=JraO!l5ht)BO(W6IB~ zj=Lo9_sXqiu-<++$S$_p#?d(C*~1EVEVhtDUpKV1Su!ti!)hX1zd_bbyuBiVE}}%a zDJb|>w&AsEb0D`itITYBi+awlGZmDU!l45Gca5CSKUpbhKI`h{W&=wmNhi0PY<24$ zYE?KfIiWDNc2TIkufJC&<4N=p7(0plz7H3Zn=8~Vef$zo8|sYt9oJ>_ijDL4pD=I| z)j|)H2iGa)Uza2L;PH5e{?=pH+r;E>dqv>NJCiQtdEG|m75tRd zL5$iP^wzGUN8lAg>&aqnACFznSBtr2E+n2-Qc++6WqN1WT;zUE8LvxHp{UQ`iJ^QgcQP%${uncuum=!T+P z3ENxB7Zlcdt3z!Y1E8{yp!GQ$sO7DP^+9Xnu^eIClE)d3@-i^)GvH@6Ev{-Y zFXSJnIo=%Og}x?cV!{*t;ubc$cgBz@BqMO`!{dR<U(l^zNTYniQ}9hNir*Wd3i1D zlgcbc6~SNNdjjC3f24PhUghSmjgGxvdWZ?27s>*V^mN>C&BcpdsUe5mMAL`h+ zpjFQzt&(_xPYerBWZu!5=wP9fj=ia3addkFj|j#fW}F0~rh96Q9O6_`vpSqAMYGqJ z-(lpe*PkylutAX~CgWpc{)7E9Buz~@-?_o=&{w~N>jVAcSe!s02U_&R*+yWdg}tn* z!M>Yh=j$P^zR#_yT$ftz`iDn4>t0s`d%t^*;Nw%{OP0!jx1GXGD$xXC5GTihErN5)n=83|ZXZpIQtF(acI~YySA;=tz}J{l{13 zWTg4X%d>-d)2`wA?$*}rWv07FpO%$fOgP;Ux~Nlc>rrWOm_EX^xa8)|YY1F`qU|s+ zX{~x!pWY(5R$6vu$Q@yR1&`Xzb27IljNc(D0RJF-;%ws!h^oP{ zSf}Y*g_oZpysNzN$!?k}04o1FeK&bzWB*-Jb+qUE*WwL{ZFYyv|li<;Iq3!d)A z#y#N|JJ*lyk3CFGRRr^fsT!ONx7ThTBM6)}D(=G&CZc@Y|k4(gKy2+YTQqF@5EqgsCG2OReFf`)n{e9$R7o1uYjo2TJ@r-BKCA{nKfU6Qvc9scbCtEp;}8~`pX(N=uLijN zqD4vHbdhUn9GJl9yOT%Rww_==ViGtWr`Vqx{-A-&jRs@=CtXU+=`*9K28jck_@);`#3 zYNjb{Zx~dR;T(6fG^5qy=)-;iPYug0w3k+&B2BA+FRO*Lx+gt2sRl+tI?g~KJb!Y8 z3%7)EgTwyN=MPw#gW~qoHzL{h>$JNICb|N}AlsRLUDE5r^MyLZRejb|IP^LbJGa~_ zB%JxOeN1zUer)|=l|mVnN33_f`ukUE_r39!=4^@GR0X4=9N1f0Th&iOwp1eI-nsQZH-l zkUHJr%)g<8#Kih<+Ny&0g=sJ!WVz4I8@F z^6zED7x4D%d8X;<=@z+_mr3v%pXrxJ(b~YK#2~-B+s(}lo2s^(MTwQZR@6K- zZ=cfDJ2E~o!Y#*JdBd;JN03Y~`-v+&^2=rJ-SLNb{E46_}LEvC8Pl@iJ{)oPv zU0TXzfkM7S1xI8`_tXbJS;4A2MtU8r9(y|KZrx;Z< zSXJW&byOY@(@9OyVcbgzZ~U3s`}!<%wnRjMoux_QlotMGjXPur-O|+stjlI;+OAc1 zG8=|R%M`TOij0fmf?QhFb!WV~MIJbkKvZKB)I zNY&u3|Kj-U3H;fW8|`EC3=GQPX79DE80e1Q#aUiK~{H{N-HQ)YR(*D6Z8Wbl?he#5V>W8uRM3#%G;@4BH?SjZlQB+Nu3 zqTk##CMK%OrN^o!$V+rKvVH6r(s8o<<00r=!>tdCm@NJa-J>Xr3q!CrMT09>PJ(4S zA3B0FpJd2Xx%WARBU!r+W+;ReNZd4 z+#K!e$l2chag7>v6~~-X+wI|Lv6Q1n!I@DX2-(jj%LYY6MBGh_2drm3CGo?z1;9!v zz zAWgnUg>Fw~J5`if3>RJ&+^P>)*&1E9Iy{!B8bP;r<%DEd6H5(aUQxjwD40$NOqRnfHbMvdy@#u)j$o z#lazt3%C?N)qE>b(O**^cGYUV+6bQAA`6S=;wcYAtzSP!!S`>bcIhuK+K zSy-Np&RGhfkSRa?{IfumARHd*`sYK?`O&A(c!yQl59CbL|Vsi5+~!{&V>NG;O@N!iHV8jzI7doK_Rxbj>fDy-KS2SDqrDv z!iyg5tSj5w&?*@9`VyziuvHZq82#nTm$$2{y5hUK;$Mest}FamSI%^Qcqs6StBICY zRF6r`AnvBPnW?Gg@%fqJ`RC=ld0V&5`cw_v?~J(i<+tdZWs4gg+`po!a6y$ zSF0W(A{*P47Zz|$gM0SSZJ`cbPfJMfS$#it!(B-OtEq`mRRvb-4yOc%7UH3n(%|{} zyoely1+A@X3sX#9n@q70aiaaDHXO3{+^Wx5BTnrDR(#!jvqBR9cCTwaX^k-WzS=YM z-h*+Q-Eo{!A37v12=mBGXys*E1yX$n)ev8tPS#4^XdC}Yl9BN5GrxwzyS!) z{w5dEWAlb42DWpATHb8PDEZfjd)Wsz%lI`^^`Ot5>u@jTD_OEEOKO-#r@2!TL|wYAiZ!RtKujlstPf3d8Kx4{d2 zpoV3j3Boln#KGSMqT-BjbYau>cDV0i?Bd`Y=J(n~0RqAHcQ>^RwA9y=b3%9vIs9!S z6zc5@dP5)z%AvjvPM$7-Y>qDX-F*~cyDtV{Z0^pAFbhe25q)1Z7dLmUaDNw*a064P za8D;$XPB}Qn?k4@7{J>l(19(~+sh|FE>scruW{wT`@dfc!`S}SB+ye4_U}bm>Kn4D zA^cs~B!#2|okS%>*ra8JM5RO|Bt-?-#6(1;g+(NVMWh5pMCC+ex`uI`|5U3W*4N zd;i_nzghQAqhHi*JM1UK@mrYIW@2yJ;NwV?iJ2<)f{Qa8u z?^XSK=jty0?!hk3n*IoHwttFJLw+q9*`eIGONS5Z@4qWk4BRwyc6N4k zt&NXYYcg+9I$gnLWGr?ZP1TN=YcQ$P$>Wm``Qi9FFqod<`Z%-pNaC2&CVq&K%Qwa;Eo6E!Qb9FLseYF&^>%h*=Ua1m~J40}%M5Z0F z+`rIqzlqM+wxeAt=!YpODJenXdSVqC&81)Dj6f`-HHczjt}z(Q+egCR*VlU^&kk6x zPw{r7ygx~74eA$51#QjTWU;xK^`j``X+c3jjqamAb>q~6=P%~gpICNoA_lI4w`OWy z=Wu^T?5~Z!&Er2=f1Lw1I$irBWcOeavU$4y^<$e|rb}orHJpzuLlP&f6QcFnPx~S4 z%~32voYp=W5fRaO#NmcMnS+N1--XY=nmui97O--gYo?cYv-7!&{Ig>CZ5&5 zKKW>*GKwy)?IB)c?HKAh1B+ED5|v0V?y7$K_QmO`*ZgC4MMdV-FTcLLP(E9y)UMzm zE(;${Ah4`As!FP>t4pP$Mujy}@fcToer&ZQ`?e#$+iTL{I#H`%?6cTSzHet^!?Y-! z&~dOg*J3@wjGgxTG6S(PhLujrsz31)JXft$lg>sNzL&pYJT(W%d^A1(Wf3CRNdy= z?34$I6fAoDT}WWG!uJ+mgdid!E(odlB<8H@3`%p?Z?9V8Y?;yoEw;znXCg~bx*%jogiz=l%m30hhHWnBuo@bH<3x;sKcNg z?VX*S$2*^$Dg(D>AoPn8-5Go}h3m~NE!Wo>RDy(fS5}gOR>sE0BT2-s2N*|+-jyq# z92|JOubrT^i>&n8HA1>QQ^Jofq;Z`n`q1w5@_Eot)rwp}Yx8uhp$BV!#;7zBvKhPs zSaD=lc@-2c58o+&XZH8^&%N!v@MLs&7)wtt1}w4-{n)VE6_VIU;Xui4aH|oUTbr2A zUZ{er$?l&Y{lwPxecI*oMQocVcz4OR-Enkt z(=AmgEE$-1L_|cF*8&ABWsX%bJaBJi=;+6X&)@>!@IkOM%vHqi%}Ow~?cx5{mpvNY zCUpkeDE1O?H&K4?4!*sO=Kuyr{5pMLYisM-vuAoaa>V2;`#&FY-1hSF^7VIJ$W;DQ zppr=m=2g6PE8&l;t1DV-w!tJBf$OgXgQ?y``@lpqUw-Zh;C{rvrZAn68lifuS{vIG zs+eCm^QwEnim!DFzPE9XBk9X!i$74I33HVrh`m@D!lp(&Db(jT*U=#qP9tXwBA`6E z1xv`*xb?BcS{mO$K-WG`jy&G{$XjerYzr|_B&kw=N!!jbtONS-M(9x+axCkx7ITQB zBMF@MMmr5_bf`B;PhGw(RLe2%3w{}hv0GNuqFoz6sk=J>L&%yjH#g@ao0wB4v+4Fl zMtOqBz)9ldLOja=&+E@&L^U3H^O-jnq@@w1CKneM^O=t`p{V&x$*xuVX0T3zI2lFk z2S3`yOeloxe3Cq_1MWvaK+q|0lrUipQSBuF9QPCs64|a7gNpUN|(Z* z=;`S#Hy1j%3&ePMP(cL8I2>Xdy}iBQKn2bBFEf)~QG|1Hak(yb`x(prHO#pt3*l+y zdp-V%TI4dD%LB@6bQBcEFM_>m#A~ir+h_$WMSNEVKVnUv61J_Z*kg*iMX7M9V-jzo zrUus4?pk&~e~ycbJ6CVy^uAVK2Y+VrLz6|!_tFq}Y6gSka1pTe8LoyT4V>~D(t_US z2XBS>%^Is=ZBwHtumzgR!}ZscckbNLa*9fZt92rlSx7qXcDRP@E-{O{LJ<_nhuUzA zd&aF-v9XZSdd28V8Pd1LNVlZp_ji=9R*`8m`giVhJ$WJ*^dnb^`-R_E%`6zPb(w4c z7a}`pA&r^Fokxi=gfA!fz1z=uW+o;K!!B@@orR7%uZ1Uz78Wyuw!>h6>qJYBBR%y9gLX{WLi``lvt;uu5(88U;# z$w>pW**#ZG6?Ced`%YQ2Ngc@tP8B8WlL~L(D=rp`!%m5?Z}>_vV5~dkN~KLz0)m1K zUouM$(&@#eyyh#;zWN%2@W06da$r0jscP(D`~LnuF{9M@>(?o1Y1kVO&qnq-@(bu3 zNIX{7*7CBl4N(I`iM~~{Fqq-Ji&b{7oSYnj#8`vi;A2wdOGqhuDk&TE0+Lu{@igA+ zk+2=t{8OJVe?oLlnwRFt6@mC%q7Q0mx z?-SgCQ&ep2?kaoE30TP@mU`#sEm^*X9B<#<_d?-S=Qt7DC|l+W?QCwMvMO{R3Gv|& zJ@ll>z}0KH+mXchXkpMJ>iqZq-d?>$TNzSZ;9f5sezAmmcDMf;CI!TqTW!QRU-{%} z_7A_mv+i`8uGME`sIRbEy$M?wNhYlnXQLuvRVpbgTxhkG=P_3g!Z?SCXtAOwu+Zh@ z%bU|x#2A7`VRK}GFs^@#Srd(*MT^zZr>CBLv$J&5v$$8t*v$Nv=H|&#`8u@WO|&-g z4OWGSL_+G;dwr?PyZslx)<)B^Bmd@xGaqej?I|8X>t_k4-b=k<4nDL-$oirmMTgG9 zcKzw;>6hOg+11P2`|No-&#UnUld4&5x`EWwJCAT zaD)4!E{;K>#gGWMx3>c;kCw-y=j7(TQz$k%GU5aPkTLPv*O!?ZXW2zX3^X*yuDdkq zD=RCH!wxsP>ViH8ww{5A$1E3^WgfjhN(&H@@voA+JhKj0{YX_{ln4YO!48-Rq%|Xh z*tuc8>s!X>9PT@099&#n2SsBUjI1)gv5dBnQBkF(rM!poZ;W3)U~U(=KUuQ%zOML> z&%O?~UcqDc_gn%eJD)?|<_jWH#-z!~R47hGW@{45OA4HJm->vKnRoAcd}sveU8}Gq zd?@awan4w=BvnD9S<{R>)$X(KqaU-~P$FXDow*jq*+4_XDHJ)!={!gR#?L7*eBPfr zUnBhAynZbUU|Q@A?9;T9?S+7g;~fM6nf`iFo9zo+Dhln(KX2S}($dmo%2?k92M3>L zw*>sj>cVJdvSf%jfkOo;V(m+9O^wk{aU=!l&G{6bS-0ez=s#!Q@km)D^zj}8%Ud-Y zc=~Bp9iEOt!7tUz>zrR_UvsX*H1;0m3tErysGp*RDL;Z`RsX3&%JWu8g$}Z-CGjMe4z3;+H8$WI)q3XXz%Tw(|=o?n?E{#?gO?Kba zH7OyX(`!M@!UD7U;K2isG$#h=>Os(oyzbT}K#l(d=^4G!4HlPspebQ8$zW>ZnaF@b zyEr?Km&((USV0@jn=O$D-nd$~X*Ag+errvq?bv*w_Oh%hA4lB_Y04G8so`M4kXSw~x2W!~V+FO7@sj(53M7k1a@7C!Gt;g6w@a6f)S z+)Vg03$sBqk6*j5MOq-d&8ws#>V`M9+ChlR7MAy4*8+Hm1z(Nn8B|wKZ@o7OUhPoo zX3u|rvI9L1}y6wlR0c1=o>v#@R6+n*3a$d{0s9;ww@!mmb0 zIlEeaEcHcPogY!+h5J8~vB9tCY;Pa>JR7!+NujKyc8g4L1|HIA({lhUKcmhe%F~yz zMknu0F-L#IiI5M)&^NS38kC7>>UwX4w8grM=Im%F9xqh<;$-*Tn>U2``12k2NxSd* zPptisQC2+zh{$WHmp*xMAcl1fH+)%r=`yhaI?C`S1h~NQ@o}hpXSN4TSV@iM=W|T5 zLd+kgXq#n%)eZFu_|#OmK{{{$(6_gbbd%Ev;=2&bLO$pFFImHKUc8VOO2{+(-OU;$ zjoqxy&SK9Yh^Dw@zqS`TY$Q2u zhzbZmEmw7Q?W_pYMdS$SoF|f5qj|n)e;v#4OYUTZ;aw^L7IJv87vx7hsWkeN0E71U z1<=MoJD_YlvJtPc*tx#)QuerlASHT-D^nRUjn)9bKm!jc%}@D9K~kbQzWeoBz<>ge$N{H$g2 zA|*XN;kKGcBU8rc2WhtAk`gAFpM!&N>@t$f*cq%o(sFzqB2v3X9m(RREKPKSB zdFQ`5#lAJOV&CkHxi~wUKD%L;Lhi2UhQ-ds5&%+CuvM{qh>*-dBEbX|+%&j?l#q}j zWM#GQvK4TnHzFTsG0oOpB>X)*IkIwcseZ*c>gSyzuENhtGBY#j!c&V~#|tWVdt`R% z4Csg8>1k0F8{-HN=7@@Bm~~y8+v;DYRBxOQq&V*@E0;*6yg$ zFh=BwQPA!b0U@WCi|~2PE91+wD_*T(EV2}y)DBWCV~LlP*biTviS%yzqmyC-xN)Yq zu%Eg$!;^^_lP(_lWD<8JJh|`Ul1Mu4BhhsvlZ}&4F`v?5 z+bmZlIF8JUFu)YIy{emoID*L=I3*;gc1Pja=Cw3DEIf1OEmlK`e#ZJ4&xAAaNyzxH zjn+h__mPGfCFBRii2h2kqzaZ3VU_&rHS&s_6&l|hN-n*MqaJU=hegiS^5dU-XL#6W zev6QhkRTejMgyro0c$N+2osZh!CSn9q&rq(;(h-7IRjnwFfZ=DDu6sU+v*frPOV0% zh;T2RUT1S=TxJ0vv;F0Tw3&6B<54N)BH&{iR}Upc?E9@BfX=&`}}fcJ_MkRp&{ zL!CZftd^B#Y<$l9v%0NYm6TKG@K_@uxN&Na!tVx{$sUXPuhCF4e#xt=<7QIqAqo#Z z6?*>;d4d~@&s!&+Cf6_kU`OiG-EypoQ|8IhK`8m&tYeaWG1S;aI7uFw*H{ce#!H{DHj#AUhjSOnN%>w4u7XdT-A+N>S$-h|byQSw3L` zw$l5tRn-#8Ua9KLPLAQRT+8c(#`x(au71t}dM#wIhKg#wMwfP< z;^$mTT4tuS8XH!yAzxP4R2|6DCbfDu(v_}$n|v#IWeAz_8BccdCVl%U#Ml zs0QxevdO*Fk=LSw8CUDm3fYL@rr{P?Q|%^6HiX({z=->PD-oLpt5!0`TeLsI&5QAq zPWW=x(ym900r>K+we`itMOxLjeXq*ON`Q^5pLq=(fSlgVj^h~!yW=%J`gYjUKRB3X zd8KXO520K1@xuMZbzvCPV!7LoqtB)FPsKntEQ4{@ttqU~i0u08?2KN`9oJcguE=)4 z-=^Tzy=wPvW>T!+)!{@rY+PQpN*`CIcY6@k3D%$AYnnzl-t&)ksJR22d^fTWWcsVO zc7M>*=4HC2R78Aa#gd-54oxMnLU88}%<{p_4U*As7hOk_7kdI;j*TTfvGhzRa{`Wh zufvs?4(l=QpZc&^6}?hid`{tey@RXw?%mTN1ZW(ITjv*I6F^T!Mn+3pCtMQMVn`j} z5)hU8BrfGGv>H3>_TkCFdll`=qqdk)ESDd1Zg_BHwXKcKDRhr7!6IDmGWuet3*g!a zAWuACmJemoTO0#mVYGstvjfP2^EpKwwzjsMe2MXzk5M5msr0D_FNpn?`=i=|?g6Nr z^I&ah8gP_q$qGFyGI`Fumt)C85w!2J^7Dg1YPB~F>jB_*`ZD5;e)s44dLA4-z_BJj zIIUS{k5il_NnIV5U-w2|$qDkyuVN9;5;8F}>nq;M4UzO+fm5+3NW3xd1Gp04VxWRn z8)w}ZLcMaAu3)OTJpG!P$G!a2leRpgsc>@?J@11Ep& z{?tEbR*_%`{p5OMy6Uz{j^!X;ek&avoz?|rY>ZFaYP-<0mzUqIKAvNHWhj0zM?>c% zUbtoZCm~H;!^iohs1w_$tNVa6x~t64^!N8057<5@du4pf$wpU>om&-|CgOru5{U;h zM3aZ#g1?Jac$I z6zlmAcOfCz4m5=D(9lpzOS)HdWz6tGX#~q$MZIAhd6d^J#TPj_ls)9V{Cq}jk}NZs zmJ5SEv%l;HqN7S>uD4r0jk}-6EvkVuFjStY%(xQ`a%R)MlpR@JX zCaKg6bp`*J)h<`fh4!;Zi$q_Bx+KKL-FUDK^wEBO78bI0+I3}&(n~1X0PhL=&6NW- zBzfRLEuUHApbIIT$lzdMe0)3*1)zG2^%AFUQ{O$s$z!XHyf8o5r+vGogt*Fj#Fmlx zW?Gc_@HZ)n6yswR%_l17~v8p4c zsgH50b$d}RRBwS~3BOsq(8L4zAiY!ad9zsrV{=cNUM^c1WI!aQpDar3`XL| z-$NQ|KnYJ$(fGT|0I33a5T}mR07!~DrZ!*#y?Pj{Y}MuCEky~3GRx)SiNHxe)Ei;{ z;Fgp$voldoq2Lq{5P)ZymPUx{yQ--u?mZB34hy8LUr*}aj3sw z0-vgxuy9)<5iMU1`?m=ofR4JK{FZde9^#eun$N=R5%cUfYq(9Mn_h4yc>UGGii!%L z8o-I|J}oUREiBLz$uUx|(g4+#l7;2h?(RFmkJlXMb$W$s-@n%_n)ZbUEhYiOI4TyX zdOB01Tdo|lG&F>KE??nQAZAfVh>2`(Zf-79j5v^ymGa-_hy56I-+Yw>O3_ffzj-5{>g{-g`8a+St*X8g=vCpf?Di z5)`Jg*APaG02zT*`1REZlJwOF0!pr_ib|lS@69`#FwnIFUYfxvPjC?BTgQO%3`W&4P6H8U2&hLaGtgY zZW(8kO^zgpyHom(9kx!B^o1Y6sA_9#DR5Swj_&iHIRgsJcQCFl3HYC}BXLG>MTEWm z%BzPQO=eBCnarF8#`-KNdAYe%nwmLW`WU9WjSa15lE(nRgY>&R5OWRc3<|4*2rFF! zgL^<(2mCjeK2trw&u;Cw|9&(3#8+#fd&-@*!80Q%x<1Fajc zcjo>3tIJEwIm4UPFpxSx>{M4*|Gm4+%AlbanwOV%v;08?Ws#hiSoZDY`dX}M%c@_y zG&D5Hq^w}ae*iC&Elkt=%)9Yow;!Z5`c(>)scSyhi|DKKTwas7xVWV1s$6X&h2GYVk; z3JMD8+7SofSOJBaERoH*IACL=VOAY_68FU+3Zx2&uUwA(O%`q7q-oaTYqJcWJ>7g8 z6@L`_Br}uLQJt5P-(;&d?69}5k5S5tj+t3-gG<481;TlBM_ZdbUYBX@#}5V;mj2Lv z$I~cmHFpgk>w)O20#PT4PyJL9-L}hRhE_mJ#=ZxHny#4ZXi$yjxdOILyBPiQri><_ zQtaFl@bq~gk~OleF9f~ne!yV8|Fn9{j_t3&TeA2y=Z?~kk=3`PUgSzMp0;i=xRfCNGY?`XNun9|ool{L?UclgX1SGNL5d;mGXTHzT; zeDYYuzwc?sPGO5Di;OSZ?NA_4vC%e(sr=y0 znHr+XEIx+SL?pJl6QFtQRQ~2kAa}B>R|#z2mKrlOsZ`>YZOuNBrA zQxDU*zmi3XqI%rl*Fxkh?TXAys+iks;YO#d5k?{z{wY~NT2B4Q{|!H*&#udd`6}@O zSL`ttKmSiaA5lbM2dtoQEM09yF(slQBq&h%beXpZw= z2_+E5;Jrpwl9in;T7aRa#%!%INVvgA9DS(_?)8T51LfgR0eW+m(BlS0-Sd@v#nPlK zl2;W==BHjOc_2O;yFLc4H)6!15NhaUGwsHB?n@;V^lP!lrpG@eCB>}CoSHK1u9cPJ zeRaTi9t;eKWk~6YLSM$wA~i%i%4EJW(63ZH{WoUmx38cLhi$@qTjH zhwk>BkeN{YvS_v(Dlj+eXmBCS$&er@o2f#&1N};Sxf{$tHh?A?(dLQjLUBJVT)}ad zf&?edo<_F)Wq z^etQH6hZ6J{h(%3WC4+`zW%ob>IS(_gOdr+1as8b$^aUT&ZY0=y^F^I=Uu7h(=5Yf z@G%<4#CrNYt;rqGbYJr}RPt@qSd#}8T#{7c9X6d3Ae)P!$n7CjsHmWgTw`RxK$0-~ zq$Ib#9J=jISMbg!yO^t|lTjE<27~RE#MiYo%n;9_8lD2ThxH=$Ay0_x`->kp&Bwxc zZZME(n`BIuD=RB0m0bv=*7x~+#dV=ljZI5NuiB#&0Wm$AnE}s24kDq52|u9+A19Xt;+odddom8~tz}3am;XjjRr>^t2Yd&i+(tkloNdJkSo> zin+d6&SX_4NhT3EKq@IIb>A7Te}k$O!9+SbCm~3?nl7KOG=f}WdH3!<7rDb}=OWwj zCxU=#KEFu6LC3{*!94abO4Dj`OB zfh>RDc_JYN4dtj-C}mW#Qo6Ify;lHqN7#w}$0&~Xo75g3wl_C<%x5(; zHRGsOSB6JM2F^AZPX#~>aW4~TXU1A$k}G;`d@!m5Rmn+h%|F5WHCps`d)v!MlkQr^ zqU~NH1hDTMBm?B}0SLB5lAih5*)YE9UYGYK$kXqZ)JIUreYAQyKSuk)O)@7 z#GbSD4tgaOSkqLvAIF>FmI)^|T|TAA-uTmb`zN40V3KP2=0|Hs_MNSj0<5j3Gxo*gR7=(oH-YE@rj!$`HbpQswMYqnwvl<6!I!9ueQU$YiqF^I?^3XdI3xP_8Ov8cJD~(vhSqD8R33UVVXI{@UJ# zdXRsF++S&I>p#~i2`6SDBj&Z%W$rd94!E~eqpm)({>pL|o7~;mAtrI%Hg7cjtJ(>f znNq8ZCwB%4C=fh!jKOb>^aH--@1M=5l#3gjkXa_>U1VU8=vFG}NrRSlxi)FO+^ov< z^7hWVEok|;0@R4uWU)VL5D*X(6XWAgNuMcH;0BF4wSqblk - Finding Millions in Potential Savings in a Tough Retail Climate -case_study_details: - - Company: Nordstrom - - Location: Seattle, Washington - - Industry: Retail ---- - -

Challenge

- -

Nordstrom wanted to increase the efficiency and speed of its technology operations, which includes the Nordstrom.com e-commerce site. At the same time, Nordstrom Technology was looking for ways to tighten its technology operational costs.

- -

Solution

- -

After embracing a DevOps transformation and launching a continuous integration/continuous deployment (CI/CD) project four years ago, the company reduced its deployment time from three months to 30 minutes. But they wanted to go even faster across environments, so they began their cloud native journey, adopting Docker containers orchestrated with Kubernetes.

- -

Impact

- -

Nordstrom Technology developers using Kubernetes now deploy faster and can "just focus on writing applications," says Dhawal Patel, a senior engineer on the team building a Kubernetes enterprise platform for Nordstrom. Furthermore, the team has increased Ops efficiency, improving CPU utilization from 5x to 12x depending on the workload. "We run thousands of virtual machines (VMs), but aren't effectively using all those resources," says Patel. "With Kubernetes, without even trying to make our cluster efficient, we are currently at a 10x increase."

- -{{< case-studies/quote author="Dhawal Patel, senior engineer at Nordstrom" >}} -"We are always looking for ways to optimize and provide more value through technology. With Kubernetes we are showcasing two types of efficiency that we can bring: Dev efficiency and Ops efficiency. It's a win-win." -{{< /case-studies/quote >}} - -

When Dhawal Patel joined Nordstrom five years ago as an application developer for the retailer's website, he realized there was an opportunity to help speed up development cycles.

- -

In those early DevOps days, Nordstrom Technology still followed a traditional model of silo teams and functions. "As a developer, I was spending more time fixing environments than writing code and adding value to business," Patel says. "I was passionate about that—so I was given the opportunity to help fix it."

- -

The company was eager to move faster, too, and in 2013 launched the first continuous integration/continuous deployment (CI/CD) project. That project was the first step in Nordstrom's cloud native journey.

- -

Dev and Ops team members built a CI/CD pipeline, working with the company's servers on premise. The team chose Chef, and wrote cookbooks that automated virtual IP creation, servers, and load balancing. "After we completed the project, deployment went from three months to 30 minutes," says Patel. "We still had multiple environments—dev, test, staging, then production—so with each environment running the Chef cookbooks, it took 30 minutes. It was a huge achievement at that point."

- -

But new environments still took too long to turn up, so the next step was working in the cloud. Today, Nordstrom Technology has built an enterprise platform that allows the company's 1,500 developers to deploy applications running as Docker containers in the cloud, orchestrated with Kubernetes.

- -{{< case-studies/quote image="/images/case-studies/nordstrom/banner3.jpg" >}} -"We made a bet that Kubernetes was going to take off, informed by early indicators of community support and project velocity, so we rebuilt our system with Kubernetes at the core," -{{< /case-studies/quote >}} - -

"The cloud provided faster access to resources, because it took weeks for us to get a virtual machine (VM) on premises," says Patel. "But now we can do the same thing in only five minutes."

- -

Nordstrom's first foray into scheduling containers on a cluster was a homegrown system based on CoreOS fleet. They began doing a few proofs of concept projects with that system until Kubernetes 1.0 was released when they made the switch. "We made a bet that Kubernetes was going to take off, informed by early indicators of community support and project velocity, so we rebuilt our system with Kubernetes at the core," says Marius Grigoriu, Sr. Manager of the Kubernetes team at Nordstrom.

- -

While Kubernetes is often thought as a platform for microservices, the first application to launch on Kubernetes in a critical production role at Nordstrom was Jira. "It was not the ideal microservice we were hoping to get as our first application," Patel admits, "but the team that was working on it was really passionate about Docker and Kubernetes, and they wanted to try it out. They had their application running on premises, and wanted to move it to Kubernetes."

- -

The benefits were immediate for the teams that came on board. "Teams running on our Kubernetes cluster loved the fact that they had fewer issues to worry about. They didn't need to manage infrastructure or operating systems," says Grigoriu. "Early adopters loved the declarative nature of Kubernetes. They loved the reduced surface area they had to deal with."

- -{{< case-studies/quote image="/images/case-studies/nordstrom/banner4.jpg">}} -"Teams running on our Kubernetes cluster loved the fact that they had fewer issues to worry about. They didn't need to manage infrastructure or operating systems," says Grigoriu. "Early adopters loved the declarative nature of Kubernetes. They loved the reduced surface area they had to deal with." -{{< /case-studies/quote >}} - -

To support these early adopters, Patel's team began growing the cluster and building production-grade services. "We integrated with Prometheus for monitoring, with a Grafana front end; we used Fluentd to push logs to Elasticsearch, so that gives us log aggregation," says Patel. The team also added dozens of open-source components, including CNCF projects and has made contributions to Kubernetes, Terraform, and kube2iam.

- -

There are now more than 60 development teams running Kubernetes in Nordstrom Technology, and as success stories have popped up, more teams have gotten on board. "Our initial customer base, the ones who were willing to try this out, are now going and evangelizing to the next set of users," says Patel. "One early adopter had Docker containers and he was not sure how to run it in production. We sat with him and within 15 minutes we deployed it in production. He thought it was amazing, and more people in his org started coming in."

- -

For Nordstrom Technology, going cloud-native has vastly improved development and operational efficiency. The developers using Kubernetes now deploy faster and can focus on building value in their applications. One such team started with a 25-minute merge to deploy by launching virtual machines in the cloud. Switching to Kubernetes was a 5x speedup in their process, improving their merge to deploy time to 5 minutes.

- -{{< case-studies/quote >}} -"With Kubernetes, without even trying to make our cluster efficient, we are currently at 40 percent CPU utilization—a 10x increase. we are running 2600+ customer pods that would have been 2600+ VMs if they had gone directly to the cloud. We are running them on 40 VMs now, so that's a huge reduction in operational overhead." -{{< /case-studies/quote >}} - -

Speed is great, and easily demonstrated, but perhaps the bigger impact lies in the operational efficiency. "We run thousands of VMs on AWS, and their overall average CPU utilization is about four percent," says Patel. "With Kubernetes, without even trying to make our cluster efficient, we are currently at 40 percent CPU utilization—a 10x increase. We are running 2600+ customer pods that would have been 2600+ VMs if they had gone directly to the cloud. We are running them on 40 VMs now, so that's a huge reduction in operational overhead."

- -

Nordstrom Technology is also exploring running Kubernetes on bare metal on premises. "If we can build an on-premises Kubernetes cluster," says Patel, "we could bring the power of cloud to provision resources fast on-premises. Then for the developer, their interface is Kubernetes; they might not even realize or care that their services are now deployed on premises because they're only working with Kubernetes."

- -

For that reason, Patel is eagerly following Kubernetes' development of multi-cluster capabilities. "With cluster federation, we can have our on-premise as the primary cluster and the cloud as a secondary burstable cluster," he says. "So, when there is an anniversary sale or Black Friday sale, and we need more containers - we can go to the cloud."

- -

That kind of possibility—as well as the impact that Grigoriu and Patel's team has already delivered using Kubernetes—is what led Nordstrom on its cloud native journey in the first place. "The way the retail environment is today, we are trying to build responsiveness and flexibility where we can," says Grigoriu. "Kubernetes makes it easy to: bring efficiency to both the Dev and Ops side of the equation. It's a win-win."

diff --git a/content/ko/case-studies/nordstrom/nordstrom_featured_logo.png b/content/ko/case-studies/nordstrom/nordstrom_featured_logo.png deleted file mode 100644 index a557ffa82f12e9172639a30014f60c19cee74e22..0000000000000000000000000000000000000000 GIT binary patch literal 0 HcmV?d00001 literal 7352 zcmcI}cT`hbx35?!(ouQ~B`87&A(YSrLhnVS2pAv~Aqga*O7B&A@5M$H5LBv2l`0An zL0UvXDNYM$Ix2#0!}TTf zJ=D-Hu3G+hw5h*=nS;NZgPbEsS&2`<7XeIwK@(AYz8H5b0pY6%`fFYUK%d@*fcX9z zLUdCE{YxkdeIq_K93IUF7nc@ufJ#a5$;gRAr6r``&>MV`5>Oe41RNqEEhYg)NXQ~2 zp?rUTK)`BvM<;{{QseKnfGb6i3z6u7fIxhFe8hcV;yAoB1S%&d2a%A3NJ@$UBg6=P zSR%?-3`-FBX9gsi;DC4aAiCnPe5W&_>~UU1MGzqAzqWw!(AWR>#8|@LiUL#y@kMz+ zpyCn`4CZuSe~l&(P0;^GjQ=>AVCLt6hM1rUI4`^da2`$q|1bl(`|lSz9SBH+FvPn8 zhk|lP;vBp%Xe?11sR#m|h&#GEBBUG~WC8V|#3ZDpCB>xV9Hqoi60(wj0wpBTlJ@pc zd$im?^ZX}%6&W=Rq@;wpl$@#*6sj&O3splRH6+yKG&JBaX<4;@c(t(vA`0t({%37h zVC}zok^e0(LJf~b5pj4k9M1h8D=>1w5pe_;oClwp+Ucpo#o>G+`X~oi?CI@mr$_a# zog>kBS8ueV1|Emu`%Bjd*Z&{~OimUFmsCZ|_e@pFN{{W3m1Lv>Zwwl{k$yuo!G zeax@2Y4L9HUi05sHhzJ%spWiyCOt#-J>WiAsqJvjx^7-YsJqlW~27vnXKe zt$O^`D-lLj7SNH6dOG2uUuF1q)(1E_KRjcnNiA4H67u=y5Nqv}; zk^)5!uuOj2*!VV__kq*7(C<-jk9o}9arEz z<$9TvC^O5(%ZpnNT!>^9*`D#4%MfAIntFko6Hn6D*4~6zJ~eqLGZ*lo(p>py$Mgku zMsK0wR^|J&+csXefJb_mBEN6!C!0S$Jf;`Iz505!NHU7-lH3)kQ6g1JgiA=eN8ttGQ0n4 z)w164?NAQP&!N)1S|fpNtRpyhT9-X)8rNYXxL6X+bKSbBD&MZo{z7x`-qu$a$noMK zu($AHo%knAv?V&vq!kryY<1|<#ciyvpCgmWveNFP%%R^ft zYSrZKT6v}Q@f+(V_XVGhpra=XJ*;*&{Id{u+kN2`KJ%|bfqWwM&aD+sAjp|)0k$A> zxv7&xfp}%x<_|Uzun!+1uw5=n-BCg(n=L&A?AS`SnfTqLb{Yzmd@sVYbZaqBm$EG@ z>UIBs#T$tgp*s1R0VXa4v0s)cN%2?q+K&*5bFxVf32nat-mQCs$^2?B89O(wZ@fN7 zZ#4fX4~uVW#?2`~_er$5PHhpC6HiT}by;nF`O|m(j{%PmPi)>Ek+-4YJ6XynJV?)h zyJb!ite(5@qnG2}xOJ;u5Cv6H%EUSDERP-Qj9qc)P3X*1d04_&$7kE)0?WOqNu9O+ zYN>ELRb+Dek7K{MIgfs*{PON%9Yd2Um?Y~fq~IR7B{Qld2Kmn{>6t3vb?>Y+sCT!cG-8; z5f@DVY>z5ynvI$oDqeA2=mU%vgb|o!T4UlC&NK^IWj8v7`+P0};q{0%Psz@Zi?Kn9;=r)`SINF^g zSLm(*wmBLiLtXcBQp7JX=h_g<6c&EGWehvcmU3m{;rW6JgnUa95TB&VS$$$r zSE^sl#Ob*@Q90?iRKU7O1!zIW79L0YqR-oP``a)V6)zCT29jRP1x@=M3_?1F7#}aE z>WZsCEWii@Hb!;jk_k4>dqj6XQ>x zJ~eIQFCN%-1O>Koka-dchs6o4YfaO#=C!snB4cT2^o1<@&aexIY;yQ=-$W%+iw`6` zP9d@&uQ6<%<0v9>cyMq#^%Ca{6}H*u=g80FqXSJT1M~5UTk&52y835NXOC20Pmc;s zz8KGgr=SY4)Da@aASg(w))$Dnd-^OwQGoNvNl=q75?jh zTWwor$ZIK0ErHaKco1Ra<4AfT6b z*aYo;J-ilw512*yRSYFhz(!|&e*UEG{X|Ak%r8kO8C#~jObHu8g~~+H;tM!6^_wsD z?EaXzf4JEvlo`jdtQzZEJ|!)rNL)4=7xO=>@#Y}(+$C|F=AE|Rv)j;G)FzM)JJoM5 zoi})XZ(=|WhJzKO_!m4TDsDxloV{Ep8?yzjx2d@!Cnav?HvWXsnay&g^;QAGi@|r5 z^9y-3pV=Xo)@dkPMfV*8x4d^c%sdCSatVHKTtq}9866T5Le_l1#V>Kg^icvqhum?( z%#9!>KM)tB>2xj%?1d^R&AR1;j)8`dqJvgwCnsU)5{Wc>v)PYa9Q}S%SO=7zi#ksR z6=PjRr=6fu^KkKpE=fx#g$Qe0+k~E{kqm#So@*$HWRauZ7pi0#fE0aqsXqB++|-a= z^lq2eq|FjX`rb|IL5>%uJ+^;}UTe=E_VEUj-puVvpk-?|MROzJ?nYOw*bU=R>5>g&6jgS+afjwcGE1P zGGwyO{gdQ~)`qiJkb!NYNE052*oQ8YH3chE9N+SEWv5CHyTT~xTJ%swyP}-P_`3~5 zXWhzQ%lNO>_5D%dBR9KtT><%0T(}_{MbFE_Y-PiuY7uI$23z>GyF4baN}zld5&W{D z{NYF4QV?O;Y3VUcpT{n$-i92+BW_qXt;SsuH8C-% zqH!_6T%xJt*vXx=Y7k9c1+tExI3gjof0`-zJGOqLl=rDZB zw3xWn7ta|@|9hj0{($o;N93iXS3ul&wwvk#hZ$n|S~A}iPrty`+I6(+GiuEbwe9Wg zfv++4RKf@p2I&yHps`YQA>B+Q`hC#S4dqp24*v=btGM7z%T`}cQl!~!F+7l*-=={N z42LcJ3`x_CZnwj#yQ1ltxk$eVMwR2^<2f*Ad+}F)d~c#cNLV;7DeR#|d%&xqe85GPA7&H$qB4KrK^`Pv{pBfVNIO0q=@cMj}BRlIVf- zOFX`3mbi1JUEiPQ{ZiF1Le}6BB24us!Jf;E6isIGe&DJ_d$N5jRtaBST~!L)-RL74gIXk%-3U|yFIGoX=5$Bpp45<-@I zD1(`jTG{)xHhtfV=4$eM*2u0O#5<0mQ!a6c1S_@miG_Dp-M^9?$y{_+Bdz4I4NZTO zFc~}Py|DCZb{EKi-Mzggy^UL?YZbR^*ZpfFJOdAYCFbh(P&cUhFTdj_SLMRpnny-R zXQ-Z;Hz~fa=~OW=7^$&opcP^CnyT6*`Zi?j@jcZHD)G8icK-M8Y$Ri{dZh4uHS0q0qF`8ov_$- z0IBPyi-d~xc5Tk-Bx>}nc;q@!lz;~;yxChmveVB#DhV&M7!#imWdXZBEB(7*D(!!a zqvVL2r@7a2a+~2yhPcgCFciVS2mn|1@3u0LZ*;n!#p#tfMefH?t0N!nMa%QjP*zr!mj2bcJ+i&G*siq!pm@wo z^DYpSWY?a5AbF(ZPeyC*HE%#Yb z0In$X^|4_QPm@QR|E@dL%_=!)ab$T2RQ6!CLEh`f4iL$vsXMj?M7Lf?M4T^D-~=EG zthPxE^)^bR=()y4-OL*c-nDw%!d-8HB@5=N@xqFVGs@Pt0E_iIE>({QS#jk7crzf$ zSsh492fu$N37!wo^7i)rIoqcHC?z;&klX_RYi)Y#hTQRLYu1un1;6@}bFAWbv=Xe_ z1NMM&Pg?83Sw<+|(eV4njyNhBmcli4d7pKnjY>U8Ol&xJTV0xlT4{e-pjNnlM5}Z4 zQ$uyuqV-(__r&rXeX34q%0@|x<$~hwxT*DXmz4bc%eq5MB8S;9=LeH(rSKv^FCnWc z_bp}ErLdo~CxwKBa%L{KW<^B@KJdD&2^a!UAB_^K?vrZT_kVo;d`CO}oZZ)QL#Aiv zF3~Xv_eFT$KHM6FfD>YxY_PVZdUX7*7=TMREpIfH0!7<}3m5QNUWQxs=)u-p37|AI zbq7k!kGfk0uM$E#dQ%hs0P%s)hH!l^=x2~7tUv3&QXPJYK3A7C3?RaT0nys@vrqIU z!TlgF4GoQCm6EzGT1Nn~Er`!XdL5`bIX!wyOwP^a5KQ}6{bDw5K!#iN-uwOPVq>mU z8Jd#%t4*p(wJJ4ySmQ?zIVJAC9#uX`@XZi0wg--9YZI?0V!Q<4#9u(k!dz5#=FV^K z_X;fSm=7y}R;ZS$>%V@?1#NT-@&^jJK9kkXsxgTJ_3kUq0Rl1{)^%*j? zKCC^JJK21Griq?n28LhMN)QfTZ1@env2kzty#H=5CoW$alRJFhsXR9_M1i>S5%X;* zXQ!#byqK^wGOi%}JaQl>Fl0p2Y5nCFx>=KYN9_ixb7>|m-k2vzP)}hSb{TB`>e%92 z#(+My@7~;dg~JoQpS%%NXPYUdA?4x>irfRf)`nLzG#s@Ye?B(UFoGn;u*BlS zoKe=-UB4$!x3;!EftLofhJm@8%0DTu4DD-z2Zi$SC(#VDWNxd`Y@D(?mqm+rbw00h z(BY=OF1U!1D*yhII}>^&$(Kv1Lp|BPd-$O!9!3B8-b8lTH(zAI+EgR6jVyq4%C^qg zcgHZ2_-@qiQO{fgECcz2Qnx5*|NC-<&aTm=^8}St<3`N1`DYh516$*pTa4y_Q+GWQ zy?(1AUF;_&nvQ#M`bBy=OWILbv31kl=KGlX4J0hd5t9(ilKNvIx zV@zU~ZNi>WwJe%aD6~#vTWBfNRQ3ag&^WWr869AT>t1h65#fPGKSTh)e)_CXx`->`tJ{TVEb4MXl-?TK|q`Ab7=9&)PdrFhm_=M8Q?rwH7doK2;9Df?9G8YyVU3picX52CgZBqbMn(tKYfc*r zjthCbzC|?)@#Db>mZ#xv6CvTy&UnJhS`#agpRkr}OS)l_m4`PeHiw-LROQ*xQZ9 z4Bb<$XQ5Y%`b1O0sm_<_8XwcE4=YK6Lms7SsOe<{^l)%?P5>Cb#PP#QeIwe4oEn2VFHo!}GZq4Th>MHjqQ zNl;&Na9+hXk+(n3)_VZ%NU|6G2{ar*bVY_0QXIAwX4mvOl{ZTIJ;C09Gf(9+UK0gYvsQnF;dnaGKRD0O*x`Fv<`O_-oo@9_Z0 z*WbT+SK)HnQ+G!_CZYY;InIs!o}UAe&79&08O35P6I0XS=e%7G`pF^hMzf&;a&nfb z$qe1^>?CQUF6j@k?e1H!Bp>>VC`>N;lMVEu;dry4sKkUD<9=5Y2d?>5Z4FO1$o8(5jdd8LX~{3#q7XY+t13aX+|}@~6?l(Gu=5Q86J} z+e+T&iXk<7x!EHHi29y*GEE^K zTx;8Q2=>Ib?>V`}Ay@F#qsndd<`YxrO{brEX{h0*o?nxZnRKg9lb}&fU@YK%1N5kX zcH;*CA8$BL38lULOd}`QQpvnkn>N1#AYu0c!k>)=pbp>Uf<5FDce1jY0(#DAsi~(? zI6J#Pg_A$i()^o^beVge-`--HVKTMFfdM63Ad>Po6aq==&84MX#yPKJVGrzE%CfH} z7RFq;ZYD97hH-wS0`!dEXjmj9!$OS!FPp5kmht*Ao{(@k57CWyH%6?)aPm4pex@#J z+t!v8xbzpT=$o?5@ymszkKan{)1MlWP67?0jTQ_F5DQtOMasdqDNi`nSRiyK z2;%Zhidt9IUZ7~KccCb%GDejYrCm3mUS2By-|O=KKQ9LUeyecf4==^VGwg=O3>|al RIZodtYO5O{YgF%s{TI$2raJ%t diff --git a/content/ko/case-studies/northwestern-mutual/index.html b/content/ko/case-studies/northwestern-mutual/index.html deleted file mode 100644 index a25d25484c..0000000000 --- a/content/ko/case-studies/northwestern-mutual/index.html +++ /dev/null @@ -1,69 +0,0 @@ ---- -title: Northwestern Mutual Case Study -case_study_styles: true -cid: caseStudies - -new_case_study_styles: true -heading_background: /images/case-studies/northwestern/banner1.jpg -heading_title_logo: /images/northwestern_logo.png -subheading: > - Cloud Native at Northwestern Mutual -case_study_details: - - Company: Northwestern Mutual - - Location: Milwaukee, WI - - Industry: Insurance and Financial Services ---- - -

Challenge

- -

In the spring of 2015, Northwestern Mutual acquired a fintech startup, LearnVest, and decided to take "Northwestern Mutual's leading products and services and meld it with LearnVest's digital experience and innovative financial planning platform," says Brad Williams, Director of Engineering for Client Experience, Northwestern Mutual. The company's existing infrastructure had been optimized for batch workflows hosted on on-prem networks; deployments were very traditional, focused on following a process instead of providing deployment agility. "We had to build a platform that was elastically scalable, but also much more responsive, so we could quickly get data to the client website so our end-customers have the experience they expect," says Williams.

- -

Solution

- -

The platform team came up with a plan for using the public cloud (AWS), Docker containers, and Kubernetes for orchestration. "Kubernetes gave us that base framework so teams can be very autonomous in what they're building and deliver very quickly and frequently," says Northwestern Mutual Cloud Native Engineer Frank Greco Jr. The team also built and open-sourced Kanali, a Kubernetes-native API management tool that uses OpenTracing, Jaeger, and gRPC.

- -

Impact

- -

Before, infrastructure deployments could take weeks; now, it is done in a matter of minutes. The number of deployments has increased dramatically, from about 24 a year to over 500 in just the first 10 months of 2017. Availability has also increased: There used to be a six-hour control window for commits every Sunday morning, as well as other periods of general maintenance, during which outages could happen. "Now we have eliminated the planned outage windows," says Bryan Pfremmer, App Platform Teams Manager, Northwestern Mutual. Kanali has had an impact on the bottom line. The vendor API management product that the company previously used required 23 servers, "dedicated, to only API management," says Pfremmer. "Now it's all integrated in the existing stack and running as another deployment on Kubernetes. And that's just one environment. Between the three that we had plus the test, that's hard dollar savings."

- -{{< case-studies/quote author="Frank Greco Jr., Cloud Native Engineer at Northwestern Mutual">}} -"In a large enterprise, you're going to have people using Kubernetes, but then you're also going to have people using WAS and .NET. You may not be at a point where your whole stack can be cloud native. What if you can take your API management tool and make it cloud native, but still proxy to legacy systems? Using different pieces that are cloud native, open source and Kubernetes native, you can do pretty innovative stuff." -{{< /case-studies/quote >}} - -{{< case-studies/lead >}} -For more than 160 years, Northwestern Mutual has maintained its industry leadership in part by keeping a strong focus on risk management. -{{< /case-studies/lead >}} - -

For many years, the company took a similar approach to managing its technology and has recently undergone a digital transformation to advance the company's digital strategy - including making a lot of noise in the cloud-native world.

- -

In the spring of 2015, this insurance and financial services company acquired a fintech startup, LearnVest, and decided to take "Northwestern Mutual's leading products and services and meld it with LearnVest's digital experience and innovative financial planning platform," says Brad Williams, Director of Engineering for Client Experience, Northwestern Mutual. The company's existing infrastructure had been optimized for batch workflows hosted on an on-premise datacenter; deployments were very traditional and had to many manual steps that were error prone.

- -

In order to give the company's 4.5 million clients the digital experience they'd come to expect, says Williams, "We had to build a platform that was elastically scalable, but also much more responsive, so we could quickly get data to the client website. We essentially said, 'You build the system that you think is necessary to support a new, modern-facing one.' That's why we departed from anything legacy."

- -{{< case-studies/quote image="/images/case-studies/northwestern/banner3.jpg" >}} -"Kubernetes has definitely been the right choice for us. It gave us that base framework so teams can be autonomous in what they're building and deliver very quickly and frequently." -{{< /case-studies/quote >}} - -

Williams and the rest of the platform team decided that the first step would be to start moving from private data centers to AWS. With a new microservice architecture in mind—and the freedom to implement what was best for the organization—they began using Docker containers. After looking into the various container orchestration options, they went with Kubernetes, even though it was still in beta at the time. "There was some debate whether we should build something ourselves, or just leverage that product and evolve with it," says Northwestern Mutual Cloud Native Engineer Frank Greco Jr. "Kubernetes has definitely been the right choice for us. It gave us that base framework so teams can be autonomous in what they're building and deliver very quickly and frequently."

- -

As early adopters, the team had to do a lot of work with Ansible scripts to stand up the cluster. "We had a lot of hard security requirements given the nature of our business," explains Bryan Pfremmer, App Platform Teams Manager, Northwestern Mutual. "We found ourselves running a configuration that very few other people ever tried." The client experience group was the first to use the new platform; today, a few hundred of the company's 1,500 engineers are using it and more are eager to get on board.

- -

The results have been dramatic. Before, infrastructure deployments could take two weeks; now, it is done in a matter of minutes. Now with a focus on Infrastructure automation, and self-service, "You can take an app to production in that same day if you want to," says Pfremmer.

- -{{< case-studies/quote image="/images/case-studies/northwestern/banner4.jpg" >}} -"Now, developers have autonomy, they can use this whenever they want, however they want. It becomes more valuable the more instrumentation downstream that happens, as we mature in it." -{{< /case-studies/quote >}} - -

The process used to be so cumbersome that minor bug releases would be bundled with feature releases. With the new streamlined system enabled by Kubernetes, the number of deployments has increased from about 24 a year to more than 500 in just the first 10 months of 2017. Availability has also been improved: There used to be a six-hour control window for commits every early Sunday morning, as well as other periods of general maintenance, during which outages could happen. "Now there's no planned outage window," notes Pfremmer.

- -

Northwestern Mutual built that API management tool—called Kanali—and open sourced it in the summer of 2017. The team took on the project because it was a key capability for what they were building and prior the solution worked in an "anti-cloud native way that was different than everything else we were doing," says Greco. Now API management is just another container deployed to Kubernetes along with a separate Jaeger deployment.

- -

Now the engineers using the Kubernetes deployment platform have the added benefit of visibility in production—and autonomy. Before, a centralized team and would have to run a trace. "Now, developers have autonomy, they can use this whenever they want, however they want. It becomes more valuable the more instrumentation downstream that happens, as we mature in it." says Greco.

- -{{< case-studies/quote >}} -"We're trying to make what we're doing known so that we can find people who are like, 'Yeah, that's interesting. I want to come do it!'" -{{< /case-studies/quote >}} - -

But the team didn't stop there. "In a large enterprise, you're going to have people using Kubernetes, but then you're also going to have people using WAS and .NET," says Greco. "You may not be at a point where your whole stack can be cloud native. What if you can take your API management tool and make it cloud native, but still proxy to legacy systems? Using different pieces that are cloud native, open source and Kubernetes native, you can do pretty innovative stuff."

- -

As the team continues to improve its stack and share its Kubernetes best practices, it feels that Northwestern Mutual's reputation as a technology-first company is evolving too. "No one would think a company that's 160-plus years old is foraying this deep into the cloud and infrastructure stack," says Pfremmer. And they're hoping that means they'll be able to attract new talent. "We're trying to make what we're doing known so that we can find people who are like, 'Yeah, that's interesting. I want to come do it!'"

diff --git a/content/ko/case-studies/northwestern-mutual/northwestern_featured_logo.png b/content/ko/case-studies/northwestern-mutual/northwestern_featured_logo.png deleted file mode 100644 index 7c1422f32b86d6f3eb930b33467899db662b14e6..0000000000000000000000000000000000000000 GIT binary patch literal 0 HcmV?d00001 literal 10083 zcmb_?byQrf(3UQ zzI*R?Xa2bFy;*DKtW{^%sk47ut5(UXFf|o99IRJZNJvOH5P4~hCp!C7d@+!p-pB@N z`%i@2Sw`3St=&6kH>e{FNz%g33IDcrG+R^hgX?H*&Yo0U@7nE2z%?P zqG|4FZ7yg56c+=CxC=cAu!T880q(Xoa3>*mQQ$x13O&{TYGwxl{=wpGEeianQM$@% z0I;1S48RNGV>9RE;Q;Upf;jm&cz8MA0Ju0f`Pn&m**W;wI5>qk1cbOa0sne{Pud(U z-U(?)%l=E()0ZglgR`@}5Ieh@n;XcD8)WDBo}E)rP>`L2i=B&$?TLfU$ph{Tb!US+ z(fut!8s=o~Xld_kX$J@VB@t?7=i)31e6sYPF4)>DEB{+C-05F|dI}l4JJg9yTb40M}ChIk;e4W@emb zFu}j&{3kp=m>(=H#m&Rd&&SQlDI*}rFUc>>FU29n4d&$3xPX1L*eGIzja$a z>HZg%`@h8sfgNE`XFEqtJ3E`ddx6>qJ7+tm4|etdF!-;m^MZH*Ov+GmOZZ>S%zvfo zpPoy@94%d87P5|Zwt#=cTFCN0*ulfe2^N&(<#}@9zwqIoSd0G@zkgzR|4S_UlQHan zb(H_3HXmYdgAtBL)LZl@%-RJj>&`pV^=3Cs} zc~#9Vok?NQ=FrEC#;VDDnOAE^miGX%4D?$6^89OD_Xs>D&twkSJ0{N3SV1Gm#J@0A8vJ8tN6b+x-G)87o)+0tMB zQT`u@JCk(0JpI|Y+TFQyH`#G1ZiOaic+wBSCKQnZBb9XbWam*+Xo&DgH@$G+uyjGvbp)mK>{%T^lS ziFIH=Ir4QD$$r1=nwRRQ{(jowscVT!gMc@vshxHxZTb48k;!hDZ%Qj!X#>aZetl^;s*^y8+UN?nf8@1{*Ut07OR91) zqq}mbH#9M38{Og2#HA6y56g>ZU-rkEp&jtprH z!Wv$-!#AZPAsY|x4t)!u!0&7owWw7V$JhN|+_||5=yXsg`%RT37z1x&DkQsK+PMKrd?f(uLQH3uzBxHF|(4YT2m6P9pv9sk4xg2 zHz;9=*VGxjnHG;B5w{V@0EGp8l*m~vcNpqU9LS=I&b~L%9k%E`;=>IfLd^YzgD^f6$s7EC!e8l&^G6c?u&* z?{;62#g@K0EQ7H5A?(E;;<4QX3rxpOa7e4mAGRZuAt}ch{Z2-dk8OE~pz5RTg5311 zicV5||2}XGm(m$wnWN1BX~N)Fxot+gE?PzhYK5x+RSM;(wn4ep_U;ig&9$k7WR<*l zd_)3ghAZu%?%JTRd>a|KLg$KrgK1t{SBs_+t2+g9&1oax$6yuT^uqh4)g2^b85P8d zAS6I3s@ts@<8E$AxW9G%Mn!JRhn`Fq+7%X~V+$GRXAN;5!bZlB?Lu4Qo&V+R(38K# z#Vpaup}V(%6Dr?k-{50WD0_hVY*oQ{tRCVt!EUrOU8yCho}!Si=2bHLs)bgXC01e{ zD+O!}Ekbj3YY0I0mT4)iO4QMmf>vs|7L;p$iy0{8Yw9;d1)--})Cq$F@7KWUkOn2j z8sT)7V8a4pCrtmC0i6@k)xNfRBk@3ofxJ&lOzrsA+g(DMKHym3r-6ThkEuUIO_h=}eT*JS5YghN znJ(S45C#?B;EL=?zlgv0uNln|t-{U0iBnM1Mof$||>%SlHQ&LX-A@L(+t&z+9D)iaj zwwtn*@xYoh`#m0Z;M4((kF`#>9h$*X zxWrLpqa`}g^a7+;yUw1|vGGnvHW-wT(&bU-QpjHib(3|>k}r4a>u3AAdPg#c`^u`q zEg{h$TbEadQxq58m!~AY;s?TXSrj+kS!s9cS*p_VmyjLMnEviZ!soT+4|TaJI0Id$QO$1lGsi& zreV@YdxJdJ$;XGzZ{a}eZ`AgutWzGX}2lvdUiQX+IJG82Usds3L!@LxiK38KR;-4;SKViR;4)H9C zw>9X_&wtNu+$>wHbtvrk>Pw+~*5g*d6kVv;^X^X=*g=bULG)!%dQEH^-3!mSK;n~( z+qW!(B%{f)mH3w1)g){RArx(nYrWYK6^Ef&Z`KuF9C*4=@@_JR!05Wb!|%5!7kw}w zv;8zJw7V6W1PcJ~v)gL7*CV(uaZvozt3M3&(hPeXWaQ99k~iO)?k4lUx0&<#J(G~- zuiu65yHI6k|5#Vxd;CyAdi&0J_~3E8K6-_wXreF!q*Go0J0Mp>;IL2HX>*gmB1LC)ysGEJ{DK|H

Gda%H|g|!l%DjX=<#=S4tsgEo8jCG$HK^ zMn!b>E?jUYKoW20BAWL#CoWgdIm;iNO(8xtfQU|Q)BRj2KufCbL ztj4i&uoa%p+~SH8unNq6F|+#fE`1^U7azJHcUy5a^as`{H@m8CHI(0bxS!2t(E)F~ zX5z=0Qd`UcbIfOhRhoqjD<6FC+IgGPR#jQ1M!lq@>axu#%B{z*gW#%nmEW>hS@}IL zZkRoqRFza+ltxmVd<7CyTOGD(O&gX#J}zgQf~gVE_ZJzVwmKU95vFM)20tdrH-gEc&v-D6W+Y+E4^klq(rBv z+ZQ|eIqU1OE0R?tXX71M`y6CseN=cU1}BR(QV8N=4o zVPR#<{P;9|1$*oB?+Wzks_(ScBJXo;4CG@3zw|VJwMk!5IcaM1Fv}5S_IAXu!fSYR z?Ze`Z49xlRYjkV#V_#8V?*!wZu08!$>2`_v>xqrV7n1X}8wEq<9AQ()bQCVHf+oMU z8rB#88g&b9;$Q{g65`Q!HhDd>OFQ9itQBye9;^L9K>Smc1jKH7Clp)pFdL_gHA5JN zEAxWbtCRaNOc4DztoL|AH>+MBtz4<^Hdtc6by;TzPZ<|KrfS(XI0MGY+EE5nUnWt- zj==iHp(%2H62!gDuHyzJ7@I?wsljf4n?%IgNel$Z0(@qKO;Y(Lc`@4JQyp#$69IT` z?<_;*i{cgKUiy*@UTtI;jHC;UpB92WL&-=S-vE@Ua=Lbq$*mW1!{kEOID$54C)O8T zpQX-MTN`LqJg(ip)+t*+HSk%wYnrl(4##~D8y_JpL!&^U5YKu%jd><28sfGwOxM>t z@Mrz?kfFfc0Xeu=Njx+E+nlGgOEJ;NA9{tkL)-5Ag=vO{Sdqc~-Zd7V^qlW2=o|z+GB-j+ zW~z+_>iNPxt@t{rnHeg%$XCv<4$>0C=f3CsVCz?3>&WHfeql&(f;WhD0>^SA0!M-t z*roj=S&PBchwc04iz({-l9<)O@j0_`5YR_Rz|Hz^PI_M)mgVr=hYqw$Vn^Q}TSd}E z;MgwmHeCL*7=u&!3-8?0(0kz~h;F!on`8`QB-YOSO8+DG)Ed&spUs{L&o&-0VPUts z?{o;rqzZezsb2 zg~~rt4}7jmQH(roH89Z7Sbp4rtBPehE1njMQz$2DR<_4C^PTB2 z?H9cDDk7=@F~2nz(^T*{ILMsfF(?)>ShWyk|3mO?yvu&*1VD8DYI>wMB(QVNfypL& zm5bI3pYVkKU7%fdaBcqR2cXa`sMZ+&Y^CZr>!=knn;zv(p^NpjKCO(g^I~-DRzCP) zd@iXG#BC1gHtp_%?7&}u7GR9Z@9~QMysWMEn3&~_C=eHP?mR?SjY_*wPWnQ^mA}Z1 zx4&*EuPLV~T;dc>5q58L6g;%%x7b$OfX20?Ssw+>6Lx#h(a~x_eY1`5Z)r=t2kXmT z#E-N+kA)|U;_W+Ks(jPlBr4UfyOo~5ex@qAyJH!!EeC{VFXc!+h&FyNs%D^C;9Iy{ zXzs8Z?rVsD-b5qh$|ypQ1d>`BB9yD)DE{sW&%zL1Za6Si*!3J8%@X_KQ+JObhf$yvEgm*k(!o_#{7yA~Rl z1+{3@ucI`~!-4QBE%x@M297V&TIQ+$DG#c~bjXSw^{;C?=$>d&Z}R0VQocL!_)P2f zP|@Y=~Qsv#8k=t^t-w7QK9U){|%NHX$fy z<+2+{OLk;N19q6J%C~N!=e-mdO1g2m#Jmvp`PEr9wP51f3mXflw#iGzK#calVqO@) z`EiiA{8w7j#;R3cBMz?_LXIa$-02nSx88IH+)Ku-DZL#B2^lSBEEsNMW6>^h*A)Of zn53oGMM8cJ$k)C7a*j3rPH@*=0)t`~1#5@&Snl>%E^|{g3QyrOJkT?eh@VoG2s*3Y zPFmLEAC2Dz!hk@oB?lW2jIo$|+m>~vEk6rX*Q3w2US8sl%buoQeDV4^>cDT=8bogk zh&G~M>xz;77>a_=13viHY-wIzW=rysZ+OdD3Jt9;+^v_HrpD?j`?Wme`nq&+odB$S z{WLQS08~#joPFrW2qUy+q=~q%ek}|mEMIriMO(ITg8dE=y4@|LiyV-e5J`^sNdnWI zMsr~TX+O4IN`nw!nU;Jf18JNbYtl?`T0_4>=agVjobEEkuVhj^i0_X}z< zSaE4UsoT!}OJj6i?e{>h5&je;FNQP8hig?CY}-p#QaL@4~U>)PK< zEx&|JNlEL*G!*}wFw$R=FWO8MGG4hd4f`#|hIt%CM|&UU`w~dp^t^(BDuZH#$gX<} zUA?1R>L;G@BO4QC2|Q9Nlf=vT#J4M)15%O?$du}bXJcO2Mla>L25`#AlLvu(gK;WIi8O9i6W1e*%asshsOJ@$)h)fK%)ite}<1G%+UJfLm+;JmDwsZ=LZ zEep!SEF266d)9Ln77abq@MNUJ#v)22Yv()PTPk|r_J^4CJnywn%4ZXAwvlSPMTfrA zX)9H!PA8FK#cNH4;Gtc=qxViiq6Yh`U$QFzvs9pvh9vhPl}wTA){Q1Ve^~}E%w&^# zA8%r`EbH0ONrxAB>clEo(@;4~tb>8VUAp(o4QJ$u#rA^&^60_E11KkbRUdP?)p#Xx zMx>Z}JqFY#_kh=a9ak#R;Y)w+;5}cHLR1FXxI3xz1$<>faoM`0kgXj_-8F>2TUxo< z-(F;qo$ew*!Bo;3`z&gq;`TM}m_#`2zhOKGDKW8ko9C_1TQ zyWN#TQBA%X3ONX+-9zDT^ttt4*Dw#Q9TK@g|Ni+W>{vP1pg}ndJr*jN=1OyKzt160 zZe3zi%{cZcjnD9SA&N%FX7T&&Z0gW$t8Rq?ZIak>15Q^}nk%L8-NX-3bk%PQjlRbt z6bs+sjb|kgwdaPR0AC|dm52Vog3GVz3k`=sOKb9x(f${1AL8Lr9*!{ay1a4_Q!Fb ziLUivwzdrbW6C<=8dhZ-YKxV|C1p{;x}}jxePg_=G8ZAZkLAxiIQ&irg0!x=i^vExWRiAPTTqHaiM zZ)1*>fKD-Ch{~9J#E>gYT4RCF^DJ-i!}K;uFdCRhtLr7@l37iI+-$|H`8G98b=ZA5 z^5k=tAJ`(^1`CNoi%ae`f}!0WpGb}l>u!+TH6=cd$ zqXXs%s~YvH(j8Tq;+XV^QG)4%s2sF>sUH=Qvsqii;8kbSJb~4Iu9q2zDQNJyGbq4h zjeSHWCCl6k)2adS=>|RPmRsohCBglxG>%Xr+t+egQtEWGzp&-EP^kl|I6sM0MP0d- zC%nd}VRRSQMKp*=UXmz_@XPLNenfD$ik!|k7)5D`5La3km5~pw>%|Efc!pv<=DlB8 z6dH^;(5XDpURk5oJ)37A0;%)O|L_17!=56RY##K@yf-rBY3&{8L-?FPGli3 zgR}Cw)u_Zk3Z*-xvYKdDy4#~1cOO3dm|9ZbyF`8$bC^{)QU52flh4g-myf2-dWVYT zZY1NQ)2X06W4sW8c`l4BkpoN28EL|TYlE20Rg>2}gY1HC1G6JDx(71jfLF5wT~r|x z%!8w-0{JLqv3zN-APyn51yXOZRA^XTDLZFQ@Wy>8)O)fYjU8mf_K z32fyYU?4X0>dJf9gL@{RWP>M=Z054EvMNkBMp_C#92(g5?pNakzmNSG1E zE@L;$q1Vt~zBz^^TtGnrAEN7<8&&=eI;Y}YL0G#RR*Q4&`=Xl5`s*H-dA!dYh0Y+7kD)?_?0STl78q-W&I*Y=0Pz zpp7&O8?{6qtL+MU{cA$$S1-L=%j)Moz6U&H9>Nopej^o@VXmC}7&mi`W{OCRPA>XW znQ#t`gfN?6rrPp)nSG?6rYdbx&E`RjS-?j)N;5av0qf}+)8@qw@+)WFw#~@~r6Y2@ z1Vd06cp=E3fL5{+>+{y+oKl0-@Ns&>_&iXo*7!raiO>0-T#&cuX-4$Bk8Y+DnX;Vt zM=>18pG`F1@uXsh5A3_j*}eydxi>9KlgLD!l`_&E8BFMZcz9lW$#4_wZBO%~j_Sge z9hZFJRNCOch(@ucr?ZrRHL%CuGW1!0VA%YUlcb@5FKRpxLJ`JtGjDIk5q`hwTB2oD z`h(%)a=@XFs4L@ywv7beZNq7rdEQ!{Qn;MW0@_r|>St{UDD$QfGM7$yz%LYn4MaUQlp13Tv zpZ7YZ9X=aEGqk%u;zeR{tf2+c8cU3hri*WV9^MXV+%FS+t$PZCf;)cxIB27rRn5nt z^e}TYT07#C32O0=nN3?X1Vkv%(?5CReH;z4_rH#O+PsBp6mNgIx#T8jO>Y!!k3Glu zee_0aZr^l&(=nw$M}$dYhNaTM{ln^t75-{gb83itPn_&w?z(Ls<1noAs33V0Kl_b( zNEj-;K0Qu5oRL6=$zXg+3RCw8sVY%uHC9XEu{UaFTveTj7cu z{Wp`tC|=RVcE6X84FdEcC{gg5`QrBs^ZVqa<(k@aJXQis@M-#2Tmy@D1Rvo!3i(JA zcm{2mS;OcGBt!`x+dK!(QdWL2@XdYXAa1tJZUc>)cwKncDzS9ExX@+>kRxrU#d_pt z&t^{)8d1)aIyftPzUMMzlxsV&sPY=1yv*1%V#m#LD>)6S>nhv|6h~AiHM_>-5VGHud4+ zPdOt|kqiU1DxbZF=oR0sf=wz)cK09lMk-|$^)@~8RC{r5v{3_MZ4Jiq68f^!al{mt zRHEin+up5DXIJ>^RVW-dr-}4l*Et=$QRifT*lyi2O#?V_q`tvUEE{5Di8Xt-P?jgSlO!8S&f_(ooF@C0 ziP{f{%z4*jKFky)Iq)1snY(XG0wY$Khns+RjE59e179h|fEz$vwENk5z~m1OsEHfI z$5m;$&+xHT?0!6laNARKewq+c@UfC7$8sNg#M$I*1I-PI^tG5J@3QLo{%kF5F)^bz z-8EFK08nLY-zH`xBpNrn5$sTat>7%S9z&HJRv)6?m3eqSwZ1ufvu53vM(f0GadA3T zYQt-Lvk#-m5FjO_g4WE(mZM!LBcdB(GQ6r5zo;>Z>}p#U?Jm$kD5E|PpuOBb@9c|i zHEFDUU*|eu{A|Ju>ko{3i5ylT8kFo$>S8ySF z$V9MHH@uW6L{)zE;C=Z1)>)QitBGdh)VY&5A+h`hwzp7@`{HCmsCavMZQfW8l+@{a zf~BZ;NydrdjLE8!5+|=aB26tn4d}!A0~u4tr_JejXz$#$?nQvft=RL&H1!Tjjnb8X z+PY4Wh>Gm)dNkv8UABp4OPd)Lwl(#>>Fu|_M2Ubbod`GZSZ?2Qy>eakQ?f8G^6W5y z0Yr7WJ<0JB-JivDB-8};OAm>piTsJ;e2xCX4%Md#-=YBa^^1|;fyoC@;KEkf6<4tT~jOyFcJ%q)B^h7Gh zWg3Tn@?1~Kn?i`q=I7{|$D9un@p>!KUK_!7zPguVS8G=6T16eY?`Kwx9M)aS8r})N zP1-{*qfs2HPc?Y~+Px}@=u?z5i8zfxFfCJbbill&BIPQ0XYOn9l)4wX!#``ezXsx( ziA0@8;9dNvG!MQhe_Q}{`C=E5DfM`0gV8VMD^5puB1@rmLoYB!aVTbQzak9x!Wv** zOKyMsaLUQ`>FXvj?rx7vB2+V4CK-xk(nyq}zJ4YBCB%F{{R%%!AWU{C=GZ#ILQ=Q* zM`T=uVRKI_r_uGHOYvVC;b zZ@~e_27N#QaG_8(1i!q{K-ld93SE8awh6389@=skfy34>u#3ckk`$O59f7UhgA=}I za{&n@I<^!iCvUTq$#Jf`UY79ZS--uvngaj1=FFD9#R}n zJ9eNQ1N6)}RtxIEaZ;u#b))6AecSKEnKxDFqE9ojVC?q{5We6>6%lTIw7kbIK#AL6 z_@`vTl%IQYcdrB&P+-CEqS!f+qJ&f<-(-?PIi}6bjy-=;^c~LK>E#4K5~))%=G?I@ z{&{JgCJ?%KS_axCnT~`9S~GJeo6rYWu^c3tLpuTkUJIiGIHXEpR)G zzHO`L(FG{vuV zvxsj(E;5&?4mj-L^Y<5UxVaN=d$8QN=xWcIbV^ZQ%X#LLT6h!#W91u8=P53_Jtr-1 z;kTYEX0zOxiz(kCNtmO;4VI!N*wlem88*f{9b0&6>jghpb^NRm_SLOEh!cx0uYxux z(I$nUTn|``XCp1$mFwtJ;>v?N&ITXF=6VfKLy;6)9Y$h*c2;rOOTAe3VmhBF*J)Qr zNi!n-jV;^c<8J?SoU}u@Gl$x#(ju+b!c)F;pFDgRdetRfTdm#F&u(G-_Ar)sCEpV~^|0dpGuq$d$L_nt11 ztq!^34D0u=!@bW9!uO7_HuY6DnY)%xJCRm%1-yQKB=X0@G!T_*V7%~XG!-G-vew{| zHQb7{He|coG3Ee(ya;DL;B~yM-kd+dy&62mi^@NeWPF#Xq&&ozmO - People at Ocado Technology have been quite amazed. They ask, 'Can we do this on a Dev cluster?' and 10 minutes later we have rolled out something that is deployed across the cluster. The speed from idea to implementation to deployment is amazing. - -new_case_study_styles: true -heading_background: /images/case-studies/ocado/banner1.jpg -heading_title_logo: /images/ocado_logo.png -subheading: > - Ocado: Running Grocery Warehouses with a Cloud Native Platform -case_study_details: - - Company: Ocado Technology - - Location: Hatfield, England - - Industry: Grocery retail technology and platforms ---- - -

Challenge

- -

The world's largest online-only grocery retailer, Ocado developed the Ocado Smart Platform to manage its own operations, from websites to warehouses, and is now licensing the technology to other retailers such as Kroger. To set up the first warehouses for the platform, Ocado shifted from virtual machines and Puppet infrastructure to Docker containers, using CoreOS's fleet scheduler to provision all the services on its OpenStack-based private cloud on bare metal. As the Smart Platform grew and "fleet was going end-of-life," says Platform Engineer Mike Bryant, "we started looking for a more complete platform, with all of these disparate infrastructure services being brought together in one unified API."

- -

Solution

- -

The team decided to migrate from fleet to Kubernetes on Ocado's private cloud. The Kubernetes stack currently uses kubeadm for bootstrapping, CNI with Weave Net for networking, Prometheus Operator for monitoring, Fluentd for logging, and OpenTracing for distributed tracing. The first app on Kubernetes, a business-critical service in the warehouses, went into production in the summer of 2017, with a mass migration continuing into 2018. Hundreds of Ocado engineers working on the Smart Platform are now deploying on Kubernetes.

- -

Impact

- -

With Kubernetes, "the speed from idea to implementation to deployment is amazing," says Bryant. "I've seen features go from development to production inside of a week now. In the old world, a new application deployment could easily take over a month." And because there are no longer restrictive deployment windows in the warehouses, the rate of deployments has gone from as few as two per week to dozens per week. Ocado has also achieved cost savings because Kubernetes gives the team the ability to have more fine-grained resource allocation. Says DevOps Team Leader Kevin McCormack: "We have more confidence in the resource allocation/separation features of Kubernetes, so we have been able to migrate from around 10 fleet clusters to one Kubernetes cluster." The team also uses Prometheus and Grafana to visualize resource allocation, and makes the data available to developers. "The increased visibility offered by Prometheus means developers are more aware of what they are using and how their use impacts others, especially since we now have one shared cluster," says McCormack. "I'd estimate that we use about 15-25% less hardware resources to host the same applications in Kubernetes in our test environments."

- -{{< case-studies/quote author="Mike Bryant, Platform Engineer, Ocado" >}} -"People at Ocado Technology have been quite amazed. They ask, 'Can we do this on a Dev cluster?' and 10 minutes later we have rolled out something that is deployed across the cluster. The speed from idea to implementation to deployment is amazing." -{{< /case-studies/quote >}} - -{{< case-studies/lead >}} -When it was founded in 2000, Ocado was an online-only grocery retailer in the U.K. In the years since, it has expanded from delivering produce to families to providing technology to other grocery retailers. -{{< /case-studies/lead >}} - -

The company began developing its Ocado Smart Platform to manage its own operations, from websites to warehouses, and is now licensing the technology to other grocery chains around the world, such as Kroger. To set up the first warehouses on the platform, Ocado shifted from virtual machines and Puppet infrastructure to Docker containers, using CoreOS's fleet scheduler to provision all the services on its OpenStack-based private cloud on bare metal. As the Smart Platform grew, and "fleet was going end-of-life," says Platform Engineer Mike Bryant, "we started looking for a more complete platform, with all of these disparate infrastructure services being brought together in one unified API."

- -

Bryant had already been using Kubernetes with Code for Life, a children's education project that's part of Ocado's charity arm. "We really liked it, so we started looking at it seriously for our production workloads," says Bryant. The team that managed fleet had researched orchestration solutions and landed on Kubernetes as well. "We were looking for a platform with wide adoption, and that was where the momentum was," says DevOps Team Leader Kevin McCormack. The two paths converged, and "We didn't even go through any proof-of-concept stage. The Code for Life work served that purpose," says Bryant.

- -{{< case-studies/quote - image="/images/case-studies/ocado/banner3.jpg" - author="Kevin McCormack, DevOps Team Leader, Ocado" ->}} -"We were looking for a platform with wide adoption, and that was where the momentum was, the two paths converged, and we didn't even go through any proof-of-concept stage. The Code for Life work served that purpose," -{{< /case-studies/quote >}} - -

In the summer of 2016, the team began migrating from fleet to Kubernetes on Ocado's private cloud. The Kubernetes stack currently uses kubeadm for bootstrapping, CNI with Weave Net for networking, Prometheus Operator for monitoring, Fluentd for logging, and OpenTracing for distributed tracing.

- -

The first app on Kubernetes, a business-critical service in the warehouses, went into production a year later. Once that app was running smoothly, a mass migration continued into 2018. Hundreds of Ocado engineers working on the Smart Platform are now deploying on Kubernetes, and the platform is live in Ocado's warehouses, managing tens of thousands of orders a week. At full capacity, Ocado's latest warehouse in Erith, southeast London, will deliver more than 200,000 orders per week, making it the world's largest facility for online grocery.

- -

There are about 150 microservices now running on Kubernetes, with multiple instances of many of them. "We're not just deploying all these microservices at once. We're deploying them all for one warehouse, and then they're all being deployed again for the next warehouse, and again and again," says Bryant.

- -

The move to Kubernetes was eye-opening for many people at Ocado Technology. "In the early days of putting the platform into our test infrastructure, the technical architect asked what network performance was like on Weave Net with encryption turned on," recalls Bryant. "So we found a Docker container for iPerf, wrote a daemon set, deployed it. A few moments later, we've deployed the entire thing across this cluster. He was pretty blown away by that."

- -{{< case-studies/quote - image="/images/case-studies/ocado/banner4.jpg" - author="Mike Bryant, Platform Engineer, Ocado" ->}} -"The unified API of Kubernetes means this is all in one place, and it's one flow for approval and rollout. I've seen features go from development to production inside of a week now. In the old world, a new application deployment could easily take over a month." -{{< /case-studies/quote >}} - -

Indeed, the impact has been profound. "Prior to containerization, we had quite restrictive deployment windows in our warehouses," says Bryant. "Moving to microservices, we've been able to deploy much more frequently. We've been able to move towards continuous delivery in a number of areas. In our older warehouse, new application deployments involve talking to a bunch of different teams for different levels of the stack: from VM provisioning, to storage, to load balancers, and so on. The unified API of Kubernetes means this is all in one place, and it's one flow for approval and rollout. I've seen features go from development to production inside of a week now. In the old world, a new application deployment could easily take over a month."

- -

The rate of deployment has gone from as few as two per week to dozens per week. "With Kubernetes, some of our development teams have been able to deploy their application to production on the new platform without us noticing," says Bryant, "which means they're faster at doing what they need to do and we have less work."

- -

Ocado has also achieved cost savings because Kubernetes gives the team the ability to have more fine-grained resource allocation. "That lets us shrink quite a lot of our deployments from being per-core VM deployments to having fractions of the core," says Bryant. Adds McCormack: "We have more confidence in the resource allocation/separation features of Kubernetes, so we have been able to migrate from around 10 fleet clusters to one Kubernetes cluster. This means we use our hardware better since if we have to always have two nodes of excess capacity available in case of node failures then we only need two extra instead of 20."

- -{{< case-studies/quote author="Mike Bryant, Platform Engineer, Ocado" >}} -"CNCF have provided us with support of different technologies. We've been able to adopt those in a very easy fashion. We do like that CNCF is vendor agnostic. We're not being asked to commit to this one way of doing things. The vast diversity of viewpoints in CNCF lead to better technology." -{{< /case-studies/quote >}} - -

The team also uses Prometheus and Grafana to visualize resource allocation, and makes the data available to developers. "The increased visibility offered by Prometheus means developers are more aware of what they are using and how their use impacts others, especially since we now have one shared cluster," says McCormack. "I'd estimate that we use about 15-25% less hardware resource to host the same applications in Kubernetes in our test environments."

- -

One of the broader benefits of cloud native, says Bryant, is the unified API. "We have one method of doing our deployments that covers the wide range of things we need to do, and we can extend the API," he says. In addition to using Prometheus Operator, the Ocado team has started writing its own operators, some of which have been open sourced. Plus, "CNCF has provided us with support of these different technologies. We've been able to adopt those in a very easy fashion. We do like that CNCF is vendor agnostic. We're not being asked to commit to this one way of doing things. The vast diversity of viewpoints in the CNCF leads to better technology."

- -

Ocado's own technology, in the form of its Smart Platform, will soon be used around the world. And cloud native plays a crucial role in this global expansion. "I wouldn't have wanted to try it without Kubernetes," says Bryant. "Kubernetes has made it so much nicer, especially to have that consistent way of deploying all of the applications, then taking the same thing and being able to replicate it. It's very valuable."

diff --git a/content/ko/case-studies/ocado/ocado_featured_logo.png b/content/ko/case-studies/ocado/ocado_featured_logo.png deleted file mode 100644 index 0c2ef19ec3b0359285ba05db312d2d8b10cbd5aa..0000000000000000000000000000000000000000 GIT binary patch literal 0 HcmV?d00001 literal 8089 zcmaJmWmr_*)dCuXXkdZ7mfdd|G?}06?Us3e`ombErEM7ajFH)$X=N zH4hO=MhHDOJA|*bCmbLTbF+a1)m*IY;ks~Zn7_veTp9qt;B?eCLKtakNZPu&@LT_Z z;rDZKM_~g1(z1T;*0xS?1keU<@8}9)K5FY?20FqZ%!Xi1kfysL+`&;bz!R<)prvmc z;AATSW0sWxO8ZHo2)Muz)<8cOXIC#tKM3<*bR|*!Kf?mdz`sBcP7vmQF=eEw4ODdV zgag6+qI|Z3A|Rlc1izptNCYg%3lstgiV1+g0w7U7kf0<;TvA97`1i$(n#~htC#efn z{(COe8-&>bfpC`;5b*W&<@Xilck{Ft5R{OR5C91Y2nq3_Ao#reT@luPe6C(B|4@L! zy=*-l-4Tv%uE0MOt!>=A5fEk+r~eYc#a&bL--KPg{#F!9WdeTI?gE1RAORPbKXUyA z?S;^V|A&qL3GJot?+zExg?qVqd)lJxVaM_hGD>&<|3rTPQQSyE-E6&G;I0TYD1;d` z!VhzVNrFTalog?hVhVzyLV|)y;u505@=9XzLJCR}iV9$H&_B3pu3iXhS6lc$xQ_q9 z75#s4B^5p4)(AIGeK$Aff2^nN;D&JXa&U78Dk}bEEg+|+wXLJ;pFyra3jUj4xTm8J z9H#8)<^udnLz0gFg9{;$l7u)^R7gS|Eb^BGLgK+j&th|8{@4E6V3RQO+GC!#e-~(Br5< z<@Nm*_swyF^qYHqlyQHX?{z43 z3F)<|tF5)#(Xrm(?~c%k#UfB~ zSQC0zIEcpE>Q5eRxEld_EcgJ>w!C!dZStqbdNk1;Y6)BIdeku&XtBlgw6fV8F7cC6>X%=$5@?tDd_Tt znCgDDHjw zSqQnni&E>-=7nhCrlM zFL6BPY>-~+#d@rcpVt*Re&n^1G}WWuuN6c`>1-!AN7;_?Rq+Y#DBX(;kPvwS#r6#eeFML}L)j6px5C?zq<(5y?WVo&s`ewP!-;Jk!Kk|w(O zu{N%~JFMhqglWoYM6Q7tlzUkDjr$C-r>!sTBL1-sEgG|}S{qvLT zOUJmuVw48KFZX$6WI=+59kI0>9RgSuR+`JDtbmph>lVg`LRAWw(-&r^4yprSYM7!qN{a zQ@g$M-MyTe8CPQB++3Q=yX~_E>YA~sZ6QHYWAV9G49v(^mxNCs+Yrng@1Y4;64!f^ zCpy@V=JdYzu+wixVeHZgYnju|7DFVN58jz&+ca+^4RJc-^a$>Z1YWgk3&ei`-~3RR z(2uOcJf4$Sbnl;g)z`W7P~g;luP@v!!MG(^Oyr(_Rk^*g+ONRyDfX}T^8yG` z%4CkmkH1d*sBGu85y{-5kk9~KTGKHS@8P+ zHyqpuH;y$dk|&TUvEOL>tpMDJw2Y5R{0uYeR_33SD#yu37KX8}=xEPCNm@5zoG*w&mFZK~g2^HUyL7xh zL`vZX{8o{nE)J?a{Yfmfx4`y|d%ea?G;Wd-DWY6>PsWr_Jtgbe`J|*fjYSQaz9g3{ z2aHd&*uC)ckP9-5k0oZ`I-wv8#=jy%563T>VvKw_wYI}-UG=0yq?MeZ{+Zm9+7q;} z)TND*Z-wD;9&5p^4FgrjQ$rZgpBlrwYo5jP%-4B|@#k1N&e8cS zJ&Ap~eN~$J+IRYe38!@cs3)s;=XK8T$}C|=9|DoW7wP~l~>Xqc+K}FaTbS-sJqt7 zcqg{4)u{SqDu`yEbg3>d>flBVZsNA3_D^oAUbRmp1*z83f^l3+WlU}t2Dgg3Lv=l} zHH2P8>ufX5(j2oYeH!SDp1~B86RE{8^q8+2-0EZZGH%|})BR~ErG$pRi6ldp|G;RW zdGR=Y?CW~iI2EGEp}4Zdb@9~H_B8zFC02JS1y;6w{epsM&Llf-Xj|;J=E_<9 zOV>sw3TUN5h@y{)6*f;miMy8Ksm#&S8L>Mt;mdo@(fw}0%vFJJa1rL2Btur);@AIH8cUB&%Nib0 z?jvLZ;a|((8vCfbHPf{kn(@+#0PXzkB{8Kxe*o9O)))2#XXftVP5R!&mWvf`BmCul zr}|s9FHb_dskQs-7UUjK&(+Ml$28Lo!BMz3Yh*WKU9f4GA!yh7c?7&=`QhYkU*Y8I zGhD#_I?;!AdBE<#jLDM!!2sUXx^#7JDc728LkZG~n@O$BwwY6p6fm?Ir-$vb_0#0C z&8Vs|H?<=tIc;#Rdyd%}doFD8ggBShu(i$k^_ zVU9~bUu|4i_I~K+!(VW)H!doFJ-8QrS8t%$fcWhtRutZsO32X8qkc~J64o+61R!1; zUKgJyv!J&VuJ__2RY$uo;W)x4%04hrFO!|@-HQ%s(CtHRKUpq*0I)InCiViuz+)a+ zUAxULUPez!NZjuGJ*B8;w7gn2ys|psV;5J4y@?=a<$ zU0{288DHxK)b}35V;2)Hic6o@vI1NB1OpVCM$9&1e05Hyb7{9CV#;;W$V##UPZwFg zeg6{0hobRpX7wFimvDBRvrhv?ybC|mYdaYMC_h6Xez1s$163xAS7lkXb4=$~h$YrrHZ>yoW6T>+Du`cBq&sVR*Fd$cvwB_j z&RzCDP~X!T>;fc8PYufy?CUIE2Zi^h|Q*oF6- zGnIIu3`AYZOqt^qPebRbAhR3b^73niqPDRf&a)@0DWyBT^;mLB5F>)h_npNy5=+TP zh=Zfca1Mnpk00uZ)j)YGJaLgK@)utK8)oK%lnj7Jd@w`o&zO8 z@cM~JLnHB$O@hLuoCregjzuGtGU<==cH_^P2f#E@b%x92MgX56R#8HDI>P>}D+A%OB{ z%(Mq~-0@`Fcz9&mQv9Zn${*AZY?kA5>rtF`V`=r%Eyj248WXo^%pwNY>^l` za7<#UP@GQ{=K7CiYl;!2(WLXNosiTvdQ=2wAuxNAV9TBLSW3xMRo4lXx$#mO zJmv#2jvM(5*|vC)pi(y>-nZl4&ah{Re9xJKAM2DY`AR(EQwA{+Qzi5+O(b1K4814B z@u0Tq;cVwq*a#os-1i_m)~AW3iH7jXkM4*(XHvGFj58Bb=EUunLz!fOCL^D|EAhj| zPZX$*j_MG{V23S>M=_G1SfGYlx!hvYL-9m?Yo3cersT5bSWq~wmvbNffa<1LS7xX6 zH-YyCe_Ba|&sU|_hch77)^mplspZIeBi_62UtzLRGriIZRfahZ*?m-XZfEn5Kbg8a zA_A+3l;BU+F0MT28QAY!M`L0_KLQgO|8ZKP%cabeVQHfEM*4j6iOl*>KhYeQt7TsS zi(~G;KJ33*e%6Q-ak0g`jy4LA!=<*q9LO7^iT_bJoU7ZxP^v9Q2%)ejSuWs=MA4NFRj2^oY~OG87W9xV<9n*{_v_E7V% z2toUxUpD=j*Q8v3aIk-ngq~E^y+D<&a=#mV{`^7Ycjx8D&$BGT5npxj`q6+ja zLhi(*eBt)?7}*Q_kB77~>X&{6-F+z6dz1w+E!WPK2z32r>V_2yWmVJ42CbJ+zd4H2 ztogp_vQpnzP)iDKXf#Mj@sau(U@`gN&gbQ);NQ7dgyFUEUxck z=lParqQP}@^C}r$bXYG}8i%ZuZ<1;0{gq|D9lc~fykBvXE8C>$4BW8L9&lh1F1PkD z`o~0msPFPMdm7&H;?rgT6(5xo6ngayivYWSIa+7_MIAl?HD7R^6}hyXRY8|3k~j~q zk7Sw_o3e8q6oLI{5PSPa@)Oyar^%5Wfliwh4L>GvJ_yotq6uMmgDwh)hyf6Xu{D;U zmYwM*80gSycjH=^nz3<9nuIf}TJzIB9mVG#3V=w!V(-(6*53hF^OX2#YVnqVMejZi zmy=iYe6=MX6CaX~a>vic&YODHZ(UeeNJK0yaw6j_d8k) zJSHn0uV`LXSJN5{S)bIktg-kA<3d1o*|%Au&1AesU!4^y!h{~uS{;*J+QZB`J+a&| zZV4cr8%u81UwUX;XIc&;&+RAk2yk&7`a~pP!zm+a>=9kj{KF9ot$tiyN1ed|9go%; zVe1NpxV_qZ{6n?H6*&Po zRets*OWV72sO z76BbCU4F>c{W_?BsNiCg+LA`hhc|y`%txkqJj3G{D2@@tP8c&0rE3{<{w7z4lQ#@kRq!?-T1}=x1m?ql8A%WODSk7sdGrJHH(eB zxCVE+i(ja@x_X?ca`^UDi-gx~webOdbZO~UO_Q5SDgB^~B=@uBHYZ?#kJBwi`Y2XS zJNhBo7$eCz=I-s;YJU9E$A|2LI8Jjb5a-uRGtA}ClbEtMH>VAg+0wgF&udJ(dLAyw z+1}TkFvPA%Do%YilwL`2obL8l^0#rfcpYx(a*thOB9(A7ThpoB!$m55JQPVHz1D(@ z8Ng#$C70zBw%q&81|(LE6DEL5{{n;)^WN;Io%WJ?+_i;OgJMLadDlYchAebxs<-B- z>MCc+c)bEmE=|O*Hs+=)QuYz{TLtmw8R)dVy$Xu%PCKAv%o7JuHQTSovb7?9ExJepkrnH1-4> zOsTNpP4t{rq~c&9KmcqB@mQo{Vf5^2l;?L4xHvx6B3R@nla^|r9Ktnh%Fz~)mL}@%*aQRZZ ze8oY8i5U|@e%LknDS~W${lyDBKobKtK6q2+6M~_b^+z&LCM#L9dmpG5pTwF z>G`L1#dlapCNEBC@y=+jNJK?q{M)DPCx1!dpAy7>ba^byqC$AtS-{Op7(oIPY&0>~ z(@P^)1li##l3L>UcYHi}buy|B6Dw3FBU^t^!-4$;ruwWR&r>mmRB>wTQkw@r)i1d+ z$HB)I?Ku?Py#NM-9Xfj~-Ay~@vR;Y5wgM95S(@p7JL1Hg62JZ&$PD#(p-?D?zsg@9eK5$T|K8xbfrOUk`*fpSI``Q7^Dlb{tr#?`ac^n}f(^ATAJBI8 zSV3j*OwOU7UuX8Nduoes8!hhdbnrrw73Ks_&m*{c5H?PRO$$9f$@aSzCV=In)gbd5 zX4fxcrs4^%Xvrh>D%XguBK?;e|#5qR|vxQ zF4S4x&XS4mjk@xK+FD6Y=AEwh^b;i#0fZ>b2%;5Abi+GwR7ho*Lv_lg!A>Hb26#5l zTE{0VtIHh)Kln}VW!8LeJxA+($}PH4fo=nMN<@^1Qtb5!yjilR3$t;Uqb?N&f0|Bm z=qHcHjcb$;8++_aQCd>+_Atyb`0UE#h~sR9@q)}Ho&paqvfpLT{~N%=Q}^Yx+|hFL zLpC976mc}%4l@z+AvfUKG)^=*n;v}uS#;Y_ z5IcAYpNoHP2kkH#<7SIO65mv-e=j!!zYLQs1wRfvnM%Rl{A&tH6?rt5HrT4Qd zd2b%kB6nSTkJ~%<1!$9f9s-J|_sqNeZ$7WS{lrMt7<9dQA25uSu(K*_tu3AH$z15ZDQjS;ZxAq>m1$Xf*GHn1 zp!NEKB_{be$a+LT8~0?YDl3c?W(;NioPD`+Z8t>RJO{I`CBP@J41zzxBy!-U(=7ED z=osE;6gEoQ@AE$;t6#4jEG^Ca(Pd)v{jTT8{Vgh9=U?uk&w0zFtAm3V^;wG0NcBui zn%sAgqw1IYa1gE7-D#0Pv+H^8AmIneqms-7WtNI`gR)S^yC8=oQ{iWIMu6EV$~V(G z;VkkT3QVJp(HU0v+C__PH`Pq z3z>X1tw?nky|mHOr&xgKfW<0%1X*fbW0OqqcEWo zUH_YYwO3pio@eu@fA{ExiX4s&;aYhHf$PYU(2$#v%fQeu0UcEw^H%2_xV&m1G0q$kBzG*kOmJYLBlS7e3X4&YZ2pZtNq!#l6b1Rwu#Y|tIsOe7mNU+ z332Q%y`XL}D|a;IW9rQ`A+?$>vjQ@{9lGd+NP?O%3?m>8MM(j(50o~{bzNZ87kXV7 z5g9r2tTBeyP`lplATZ1vI6*gy53d;P?|+q^-a|)XT)oufw%QKo(J(2~l4_ttn?SdZ z3Q!;&_Z}g8z7ptqm}@L-_Z3euRK?wT@3$(96*Lr0KuvaAAZ;!3z+pd&etwo`0@xM! z?PtrR^u~HGb^;D3Au%y}K7$A_I=3mv&YoLn^h2cEOKcJ?M#(Rf(V#f;p;b$Bb3Fr{ z@TMHZ)Y?~UKp`Q0d%%o(v_{CqjsL3`L}v36H&N4xv=Zpusp)x%VS7ymsvPtevpJtw zSXh8sD&}O>y5XS#w`0ebilm^hI0gc`fv>H#%`_XrQp3WjKvK|oq^K#XV;{d+7DR%=p zp3qc|bzRqVcHaMLe|L4w@g~K4G!Koc-G1yfnul+?p}qoWozOV9a??S(a8XVNwS5|T zdU^3%ZVCF&$eO(KWY^LQIaF`O0c?ethz)ouB&SG@d$x~Jl)+XMPAL9&g*2Q^gNJq;~> zT;$9$xKOLwfd^_Zs9$eE6e4Sm6vnP)=6BA;O zYYFi15_t5nD==xl>;|~=GL7j+S0ZBw!;RN9o`?wY^Y(o8rR1$L;_bA4e?oA=pMy5ee)F(@%6Q+fn - Launching and Scaling Up Experiments, Made Simple -case_study_details: - - Company: OpenAI - - Location: San Francisco, California - - Industry: Artificial Intelligence Research ---- - -

Challenge

- -

An artificial intelligence research lab, OpenAI needed infrastructure for deep learning that would allow experiments to be run either in the cloud or in its own data center, and to easily scale. Portability, speed, and cost were the main drivers.

- -

Solution

- -

OpenAI began running Kubernetes on top of AWS in 2016, and in early 2017 migrated to Azure. OpenAI runs key experiments in fields including robotics and gaming both in Azure and in its own data centers, depending on which cluster has free capacity. "We use Kubernetes mainly as a batch scheduling system and rely on our autoscaler to dynamically scale up and down our cluster," says Christopher Berner, Head of Infrastructure. "This lets us significantly reduce costs for idle nodes, while still providing low latency and rapid iteration."

- -

Impact

- -

The company has benefited from greater portability: "Because Kubernetes provides a consistent API, we can move our research experiments very easily between clusters," says Berner. Being able to use its own data centers when appropriate is "lowering costs and providing us access to hardware that we wouldn't necessarily have access to in the cloud," he adds. "As long as the utilization is high, the costs are much lower there." Launching experiments also takes far less time: "One of our researchers who is working on a new distributed training system has been able to get his experiment running in two or three days. In a week or two he scaled it out to hundreds of GPUs. Previously, that would have easily been a couple of months of work."

- -{{< case-studies/quote >}} - -
-Check out "Building the Infrastructure that Powers the Future of AI" presented by Vicki Cheung, Member of Technical Staff & Jonas Schneider, Member of Technical Staff at OpenAI from KubeCon/CloudNativeCon Europe 2017. -{{< /case-studies/quote >}} - -{{< case-studies/lead >}} -From experiments in robotics to old-school video game play research, OpenAI's work in artificial intelligence technology is meant to be shared. -{{< /case-studies/lead >}} - -

With a mission to ensure powerful AI systems are safe, OpenAI cares deeply about open source—both benefiting from it and contributing safety technology into it. "The research that we do, we want to spread it as widely as possible so everyone can benefit," says OpenAI's Head of Infrastructure Christopher Berner. The lab's philosophy—as well as its particular needs—lent itself to embracing an open source, cloud native strategy for its deep learning infrastructure.

- -

OpenAI started running Kubernetes on top of AWS in 2016, and a year later, migrated the Kubernetes clusters to Azure. "We probably use Kubernetes differently from a lot of people," says Berner. "We use it for batch scheduling and as a workload manager for the cluster. It's a way of coordinating a large number of containers that are all connected together. We rely on our autoscaler to dynamically scale up and down our cluster. This lets us significantly reduce costs for idle nodes, while still providing low latency and rapid iteration."

- -

In the past year, Berner has overseen the launch of several Kubernetes clusters in OpenAI's own data centers. "We run them in a hybrid model where the control planes—the Kubernetes API servers, etcd and everything—are all in Azure, and then all of the Kubernetes nodes are in our own data center," says Berner. "The cloud is really convenient for managing etcd and all of the masters, and having backups and spinning up new nodes if anything breaks. This model allows us to take advantage of lower costs and have the availability of more specialized hardware in our own data center."

- -{{< case-studies/quote image="/images/case-studies/openAI/banner3.jpg" >}} -OpenAI's experiments take advantage of Kubernetes' benefits, including portability. "Because Kubernetes provides a consistent API, we can move our research experiments very easily between clusters..." -{{< /case-studies/quote >}} - -

Different teams at OpenAI currently run a couple dozen projects. While the largest-scale workloads manage bare cloud VMs directly, most of OpenAI's experiments take advantage of Kubernetes' benefits, including portability. "Because Kubernetes provides a consistent API, we can move our research experiments very easily between clusters," says Berner. The on-prem clusters are generally "used for workloads where you need lots of GPUs, something like training an ImageNet model. Anything that's CPU heavy, that's run in the cloud. But we also have a number of teams that run their experiments both in Azure and in our own data centers, just depending on which cluster has free capacity, and that's hugely valuable."

- -

Berner has made the Kubernetes clusters available to all OpenAI teams to use if it's a good fit. "I've worked a lot with our games team, which at the moment is doing research on classic console games," he says. "They had been running a bunch of their experiments on our dev servers, and they had been trying out Google cloud, managing their own VMs. We got them to try out our first on-prem Kubernetes cluster, and that was really successful. They've now moved over completely to it, and it has allowed them to scale up their experiments by 10x, and do that without needing to invest significant engineering time to figure out how to manage more machines. A lot of people are now following the same path."

- -{{< case-studies/quote image="/images/case-studies/openAI/banner4.jpg" >}} -"One of our researchers who is working on a new distributed training system has been able to get his experiment running in two or three days," says Berner. "In a week or two he scaled it out to hundreds of GPUs. Previously, that would have easily been a couple of months of work." -{{< /case-studies/quote >}} - -

That path has been simplified by frameworks and tools that two of OpenAI's teams have developed to handle interaction with Kubernetes. "You can just write some Python code, fill out a bit of configuration with exactly how many machines you need and which types, and then it will prepare all of those specifications and send it to the Kube cluster so that it gets launched there," says Berner. "And it also provides a bit of extra monitoring and better tooling that's designed specifically for these machine learning projects."

- -

The impact that Kubernetes has had at OpenAI is impressive. With Kubernetes, the frameworks and tooling, including the autoscaler, in place, launching experiments takes far less time. "One of our researchers who is working on a new distributed training system has been able to get his experiment running in two or three days," says Berner. "In a week or two he scaled it out to hundreds of GPUs. Previously, that would have easily been a couple of months of work."

- -

Plus, the flexibility they now have to use their on-prem Kubernetes cluster when appropriate is "lowering costs and providing us access to hardware that we wouldn't necessarily have access to in the cloud," he says. "As long as the utilization is high, the costs are much lower in our data center. To an extent, you can also customize your hardware to exactly what you need."

- -{{< case-studies/quote author="CHRISTOPHER BERNER, HEAD OF INFRASTRUCTURE FOR OPENAI" >}} -"Research teams can now take advantage of the frameworks we've built on top of Kubernetes, which make it easy to launch experiments, scale them by 10x or 50x, and take little effort to manage." -{{< /case-studies/quote >}} - -

OpenAI is also benefiting from other technologies in the CNCF cloud-native ecosystem. gRPC is used by many of its systems for communications between different services, and Prometheus is in place "as a debugging tool if things go wrong," says Berner. "We actually haven't had any real problems in our Kubernetes clusters recently, so I don't think anyone has looked at our Prometheus monitoring in a while. If something breaks, it will be there."

- -

One of the things Berner continues to focus on is Kubernetes' ability to scale, which is essential to deep learning experiments. OpenAI has been able to push one of its Kubernetes clusters on Azure up to more than 2,500 nodes. "I think we'll probably hit the 5,000-machine number that Kubernetes has been tested at before too long," says Berner, adding, "We're definitely hiring if you're excited about working on these things!"

diff --git a/content/ko/case-studies/openAI/openai_featured.png b/content/ko/case-studies/openAI/openai_featured.png deleted file mode 100644 index b2b667c0bb13d09866a2bcdcf39c77f24a55c29d..0000000000000000000000000000000000000000 GIT binary patch literal 0 HcmV?d00001 literal 12132 zcmaKSWmH_vvM9k4g1h_R?(TyJch|u!xD(tpIKe#(?oM#m;O-vW?U8fNci;PQ-kr5( zcJ1C}U0pR@UA4m%7v8^qpCI3VA$2lN-Y>Y$;+oDX_GZp* zMvfo|5mS3(5Q(&{kvT{iWMt~`bqK@{0RiP`sjBI$Dfflf#NL+C=r0bVyRE}JH3S5| zkh_DCi8aWX#2935X(vE-*4jx%VreQsrokb{EaxBwvapo$bOfn*%Bz}qTAT2gk_ict z@VoQAGq43Y8!o|bL!pY3e!9q{M%FM#W#LU6O%*nva!pqFf%gRFXpAXr)G)Gf2US*)fe`LMi z36NPhJ3H_)F}b<9F}kra+B=#vvGDNlFfp?-v9dC}V=y>**f|@yGuSzi|HA(a6}|#aV#tUDJP0!PY@e?td8DIsL~_?yh z&eA{uviB#9rk19>>?S7M@8&aNVCLjxWnkwqWoIyA=4O32ATuk7)!3ND7{v1rpZ`V= zaEWjMxkQ;o*~GZG*f=Eqp_R6CayGIv0sSM}@?G|SXj%VP zT3#_nkdd>!qpH2V%|BkCXkqVc?_^=`Kq4mgchxx zbOo78INIBi{L9w7mj6YMcg4kcL^#;FIN8|#`yBpFYx>{p_itK`|1T}myE06F9p(RV zoPUPiz40&jKVk5G^FM(EvU`sn$M-M^Li@k~0YMig4HQvzUpmqIsH;Bv)VaU%z}V2# z=U=h`i6*YJLMJyCpKjJO{bhN^0F->H*Dz8;P|CyBhe%68EA|0506K3&xWDy`_swZt z17k7urg81*=H=RByZvs(YvLtq-=kwftIMttomTbP59+5{jwnUmXG~gDwH5iC-?SF| zV6)?FEtoW#MXdZqv)WpMva88Pn|XfbrpS!Y3~eD=)Hi*SAo=3m{xgAO*Avy->KQ-* zb=WbYbWoH~9^n4&0B3TfQ^C~X>^zHItc=4I9pa~zUDxMy@Ot}|%I~FiJc$V<0uoaZ zERa_BGdYQ=w6G{1L@l&7O^zJU@LPPQgq<`;>guFYR>ZeSWAF2WA?o@>~AJ zr2mlB+eQwtmrXwuRM4B!lt2n+N_`u1vPM;-@V5EV<}EObr`{*iQ-xEW-F6QedSFc; z83~?tn`>JmXRrEIt6g^s10f8T4RAz|9aN-@i!yyihlZ$SAs2iDI^V7#K0%%x2pWxN zjxDpRlxb5K`-Cle9DGssml#L~?_L>rREc02pFO{{(z%tdS8gLfKuU+GXW)L2&gWug z4_vftTmqY+!{NqkM_kFkvzkoDf=0b|U%MvltZd6Fnq3!cZ{Gy&J&io8$NFQlIGkVU zkbgT>(T`XQ8nj-7o`zAmfaQS!V^&d3bX((E`Oi>%tn?~rAINOfmrHlRC9_^=x0FuW zcqk=#EBtiZhCqHL;?;cBZccJKU(u< zk3Nu8mk=?toUDnnrSo|^pZ1X_rX;g``Oj(?>@$RzVWn~A7*B!A7fSfbQ~66HncKai=Ur`@UW={NM=4o}A$dUUacBFuo1fMeGrQHO{qWTUuMa z_e{~xvimMMOE9O)sjbY!{1ZQDjbG_8eX*lgLeOLFb@tXKdR676KZ;{w}-9;?>a+} zd>`U|xu790M!K>*^g&PQaDSa1VDeUT-_k8@KB>oKT18!4PBI9RjqUg)vi79l<4M{Y zH0yx1W|d@vl0gAPXBBaBuGimg%zij{84xn}h>n?AI=Q|5<$GqY%XxKS=~6vu)$qAZ zabng-hXbF(B`mt}kXvu`)?Kn+Np8J~2o0H{QmH7qmS>9tK|5|4;F*j~RmE#wO(fFAUYHiQJF%NNSit5qD zZeZCxAdvu<{CwT+?fw5aY6q(Zh z-{z`dwGzntZ+qhNWPlo9$IPeS3TR|d;th%$>YdHLc^U%?W)HHmU$iZyOAmkGqUFpG z`qpNrEt0b{3(#z*#QMqcxYP1bmU{ua0t#d6EZ2W}{z6${VQzBkBw9#1TXx=X6e%s9 zTXkN9&e`ghC+T)UG$sTl0uDPLyHX+;;{gm(qOIfg`!c2*?2#o!U?>cwA3h>m z!Q;7v5)(Kq7GM0RlDw|rv~RcRLi4ur)z9Q&M|@4zo&u>tVzYR=XNn&oG$V?N##_Da zcXIu}YaR!8_cg{JjQL~-Yp7vVY2A{-xB%R*W#RZ)*eE1GK*Qlei}b#6fbrQ%b4s>f zC6b#E`>&aS{G&>Q5rhmBT)e5{MtiRPA`F{HXRqtz9yaV%HbZBjpAu{M*?h<0xB`zb zFSlcJCr$hjbd_o%JJV)wkEt^doOW#eJlqttlDwHpG(^zhv)S>;g^jg~_{(t>aDbx&X@X zn~0g{eA+>NN_h&dV;rZ`zI!dwAl7z0kL{U_vEN@eTk*+1Rp%S9nu#%MHq&0-4Lg{b z_L^*R0IO4{^H)(6Eme+b-?uei*3YeK%cvjD?00yYJw}>O{80G`C;E}^+cfTKn3U=v z0zk^VZ@$lz5~RVGvmRH)<>v`H>FM##c*x>WXZHK8lHK+!{_C(7AU&eV`8|5* z#^`(G7#`SjiWHq3S0M6jkQH3IqI(MMrqHW{+7v~X<0zP_*sI}d@wtw|kl)J`7hMI8HPKvK`B^vDK7-za z@y0ijlTsS#BI=_r>b`-y*ru5v5|!4%Ug+|kf`-!-^sUlswut2{JfqOL=v+!cIkB;0NLcil|vVv6iOJe4Pz1A=&bUOV?Z zCOQvl2C7I24)+DQn>|l|)^J{)A~Jcb9X~sM4wZ``VCNOh-~KVMK;Mumgmkt{km+-R z(8G*TfC6~~cZ^r1V+(MdiWp0L-xpC>6-*81qYxA6$>>F|wHpPjkthkI2n0cciJ_Fg z36OKy4k%as%-4UBu6Mqe3};1w?|%qb-wB*g`>Y1EniiQ>-W$#y2I=aio6{j9EA%wa zU>l)_GXD@v3XRUACwH?)P714l1mjP~n$@y1j3h|1G11w#I#vTGnlC6WLnb(j2?Tkp z9;-;V$qC8P?1am184MmC3&DveHwL!S0tpT7x61A6_8f~8N_JC2ZhvH*{@@UJ3szdJ zg;8B=i*27tq7_E<5+vl7Q(C9w$iMpvOdbi*3i?DmUS#3(O%gmWjB8YM#B!6dmy$$v z@+@XS=F|0}{Z?Mg%G+rBa1f>9dp?M_0(ws$sEk$W>ej64o*q4-(^wpPf9GnqhbXpu z0*NQH_x0%0wkk zTTN2t@a&V%qJkSDFUw!Q87Ws9%0?lIM6}{FKEuZ6Y;C~Cbdb6Whv5E6VcD&c&;4*E z3GS+Y?SY)`))%YvTc&#L=ItV=6x1U-9>xJ4ypOT3c>uJk!4(E!v_L+{C=`-Y=F z4-9p73=wi0^T_ImiyNZj>TFkB)SB00ZLbd%V2IaaEhtUlOXdDX!WJE8M!U;XvP*D= zWIlSQ*Hh^9vvSN#B&(9qPaIB{uG%vJB@PibzGu`aN+|GGO^f z1zoVFk`7*SzCLUL4k?rUe4~MzDR7{p{A26QA=XN#hTcc7#`B-Y#iNO%H|ushEnJH^ zQ*7XIo0>S-s{1mkk1>A?ZkYYCJu-yjQOKyZOk*Zd=vlSOeOI*wZ|1s@{7>QLLu8BX zJwxw4;jghV2!=Anclwy@)({(1ggS62c{bY|Pb}ZP;F)km+Gz)_G7i>pziy(kVav@Q zu+Jim%kzvapBCtolLJV<$XFNu@x5CiWO43BX_wuzR-N~TinxqS5xn#-Z2~6xd%>?J zoMF$OSXaRLV{^XM*$z}Hf%>o*^1nk5*>Ke#?7ZA6$Y#`oeP$xC&Ds*aD+^mm{j$U7 z7HoSf@fCP-cBAZ4D*>uj*>C%g2|fEZm;6B4G0SYDToYn6c7DU3izrV|N9|3#|DNHu zlI6**QoIWYsyaEhJAPdWRe%n0E~9hN^enZJ-|5lrH>}Yi*16nB^^$11)b&gXb1{B8 z%Jc&-skyFwLN%&Ot z&nrIu`N3jur5EYSdvUnMglxk*Slz-Y!jR(g##+4ln^z-TlTSIe$meM-c3*|bN;4V) zpv#b&$Syk#u5qQ0c>}u=4GQ*H*N9eUs^xu-=p{ym*QM3=|atizE3f=}k{kQzWYu z(=rB_rL=|>HLv^0<8N-^M{3%(+Or#d>cx%=MIbz%+?ie>w(iH}QgFW4UG{Kbieo+VB@cD@3bJ9z9Seb(irYmNb(>XN{A3YgD5EvL;K@Tl zZ_LhJTj$X*HKfHcRwhXNUQXNyh5!|dX5GWsZ@pfUU) zQ|aro-FYwJiClEpM$7dj(8u2;z~_BMb>*yN_hI)w3zkib>AWc`oaDoWa-UE}U+qBk za^j>g+E>swE6ftb@DkBr8Q|YS*1zEY6t&(xH}`LW>m5(^FN~%X{4bd7r{Z7uFTpF> z>#cYyQ2)IOFf2IG(+ZP& z?#mnp$f1dT1nWDdzK#zr`zx^55n*hykq?#S4ap{4K1!a)g&ke$TM&Q@1VQ5eEh=#! z4lbTVDcgutj0zC}3AsZpc1M|dn90}TKA$(bAt9`4)NB#uE-+NVgv**I+koBe9x_lK z83nkpG`(IjY5-y_z$olHH_`(t(ovOw~B#%a$L? zWn5h1=VB>ZyZ!h8HgkNfPj(LyWs6qT%V71$OQqLvPXGACRo_ZH_-96O(FYNcT_vo@ zRl8Qgz6fe5p^mqa!kHAFaMsKa;h>}Gma>gfm%)F(6Qt&=(|0lpdi|cRJhM97@25Z z(-WO~CZ8>L=N)gVG2_5Ld=L{0ePzZ|+E|Yf69R_c1_4YP{iZHg4QITekZ{EdhybL=aq_JZ^CbZar3yzNNIR|9_T-7 zK;iG2BNECm?k6R zN2pnT|LEz!S@+0%?GmwzCj%uHpd#d30?AXu&DXnUd@g||P{pEI^?*d*aUCqkOA{y( zUbiqN+o7YI89WPV^yz(Puv-d{R<8|C#-U#UDmNu+)L6ghCt2NWyW2RbD|0reON9{R zr}4aHd1a?El2ANSQuEl|#do)LXE5sai{v0+s)f6-$B}#>EXzZEgCrvg7fkoyho|{k zS+mU?!%gIIlbd|1&AGSom2Kb_mx%1V?aFj84wpvj{T#lki+Jfuv|SlVMDQ~rAWXLB ztv%SNEa)r_8(KeZHv{>$8Y_I=MRjB&4t9yc>;6D+p|QbYb)_}p2#^Gn6eBVxp6R>1F=Zg z7`4R6y!ci}n$q&s{p#Y2ER!i!kP2fu&&E4zUR8GOY~O-h+c0~rPPK2 zb?~AVrWW5HWgSwdIdg9MgqXyXJtCVZoo22ul{GxX-m9r9-}NV}J)ys1$Y|Om%TcLs zrYPyU8?9>KBijfT7-_-Qc*-zsHc3N2%@apx*Yh#cGhbG`iculho%=Gw+}TA4wj8@J z#XCpzAAQdwkyAxO2!c!c{;xZis^y2l$u>HH~CmL?E?#zMCA@+I^5+*}9qUdLVzQ20qOpP`)g|1{OX=;68icE^oo9%e1 zZlFkIFSLwoleX7Arjhb{H{hAfSKQ8OO^j((Vaf%MZS=BI|K)WsD<7|1qLOU`MMRmu zRxbhcNjkk3W11Bq-j{v*;C0))_H(fSZ}Z;K;(fynufTxS57!`!bCX{yD&`%&FQM?} zj8-l`YE|G+$@7G707#H#rU=?5W^$U0L`*z<3vdrYh2d5 zVn-IM()mFwXlll0c?GNf>Cb(&sDS@l-R zkY(LEqYleih4{WaW-Y+F@x0-4r&&z)akJuT(BV z1Ub+2i*=>p%#A*+^~BK**N&Z2q)HolEh2lmnxLlVU{JUS_veP`revhh0A(`^=FR?L zK+x*I*DHt^j=KIezP%-dCflIs1Ct~Np19nD>mEz9{09S(UL6zEYS{n{p6J!08vR1` z^&Rm=Z*9+!l8A1UBx4R2oAAK(55goy>nsv`;F>TBhw>EoEIUUvMwLe3( zwz`vmrs2*>FF2}qG;31;4_uPiq#819xDO9V*j$(l+rUp;kD3Z?q;y~p&H0g%<@ao; zypsu~7aS(lT>+tXn@C4E%vH@>C=tt%3)J8_+V7#%8=#hG`00)z`$p(`kWQA)NCht5NC+Ybmv)DUOhTZT6H=Z4n^KC zQH58e9E;q;d$K>_D<_mHA76k53(-{_a$p;Z4dT`=0OX71m zKKNr42M+l1AtAqN4Ue5OuhjTp{U|*i?Rrq^uo^n4)NS%w6yw5cwv+B9)lm-)N%at# zBI|&Sb|*fCsARai%ucC?|NaF0wN@g6Q}+{%NhIP>LF$f*H12H7O0zFlu@P#(f}D2w zvp}qTX@9r1!xbz4?Gk^+Y|*tRDz(t5u9>5mPqQ)PrMmC*N^6K5k!N)9xMGdVfO>2v zOQ(qxm3g@lR{;6a0bwZKbpcHAME@Btk`#p!WrTO7h(8DV??5V~ZliCPgvyex6sdId z6=>vh6*`qiQ4TlkB78wC2 z+!A8LQ@$uUO7CsMqvi@fMXHeeOj2mE>SjyxqwkvuI6TdM2RR$sezUlBCVz14>K7|G zv;k-ujLCBbVjthaW|`^Dn0h&w!jNZ@8V-U_q@bE5dkEg>lAf6Ij(s!sn^3Xg5>s-y1!GI1Tv}N_tcgC@l`_xg$i5U)@OMHOW6#<8MvGZ?^m3DLU}#d!$RAO%yVmfg zbWpsD1C1R~2w2Sp)4-jwY6OzRKZPwyi3U+nK<}~zw&l|~IH&&EC7T<&$GAZh5e!nY z57GoeaX&J?79HLS!>XN#e(z6u=?up-4;`*g>>X<{H;0>7sNm{VFPThbxmfGaA2xo2 z;Q8*h@JG$%aiO1kusn<$NE=y;jjfx5^K-5XVjUR?MZ6ZmpBw{rnAu4Lg~Mf?T|+e$ zXI240>&*r8Tm)-N%ZOhCtb~V}3>BOEBlD zy|YGcvy5SH6144+Qlc?!9DAT>6iAMX^4wu_5k!drxT$cztl{~MWvD^M^Gr`?aPd6; z+s@vhS%-1I+n4x_dw&xTt%poVKK9q~9okxru+GlThdC6bLm!Q4u6a?3=4<7}df#G+ zEnQzy@dWuzFmwh+N$4<=`O@G$rdrx()y%MY&)XzB+q7=EK?@Ej)$a-$A@lw}et$5_ zkp5W&rHC0Df@kFZgP|}NE4bW!dJ`iV6cn5Pwg86OL zMkS<`))s#Xq5bXdfD#3j$Nu;+F0|8YIZsCnxN;8wKn{>Fe<4Ft)Ld~nSn7}t1`rE; zuq?Jy3~`;oLmRNVX%m~|&fbipJn&qJ{>hV3wo9ERD;?3q<~=&9grv1ZeHKnpn^Byb zHRo8i_PuoFM-`RGz!8Xq4?v0m&RYeLm{*+~$aQ)5&YIIT@Tzh*1T7fy251RSvbUSK z{2qn5JFxS`&+~&q&1k$l#lv%8tX-JGph!1nO<)~RJA|@Dfn?e%&MYLM;rqeISqasD zOO79A8MoTcrCC!wDkZ8ae%t+XE_4@XC$38-JDA$5{@P-_p784vDGg_EqV&hzoZ3Zd z8Y-Cw*orHoh!V#)UzVYwf*|**6=yADama4!(_@;owi$(S{PwtoD z;OYL9IJ6XPZC!cDgA(pVv&Ikvgo<@H;ZIJ}PI&Y@$2*qkrUcjhjg%(&IxhUFp|Pz& zCffU&Fc5UZUm7=MyIt=Lf0JtW48}AMBm$Sf{79eOpIIV@2Od@di(|tB7vfc#qr(lx zUAZqPznHIo5r$@TAxP$lz$R7a(Z9jdll321lDmSJ^38DuPBZ~ayLVYn?+Lzvb(At) zvjga1#J~w}eA5j=9dWfMQ1mF>WaGL<*o^(!|2?$Rm{Tj319iS6t4Ol3pb~1!B%mCy z0qTwzJh<*CX>|8%!rsIb#hbCkMZmnbC$7RG@MN z?j@Yp*2`eglyQCimnq(p9EbV@gv_)AJ#FQ^#lZza)uNC$@k9Km2Et{83^d%%8=N8G zQK|@1r-TWxUWadgx8;q~l&X>M{!pxYzSP&QQV9l5>z$@gE&fKc5OF|x$Jdy8B^muj zB%r;3Gs{~|NFPQDby|(gURNBW8uCV+je}|IHF+8e?G>ARZ!rj9e*5WIeTgszTgas1X2?ia|hgb1GA zXhrDN?p6dGbp1|M2TnDH;?;WBw@tmT^+YW`a=*{$+-eA@XmEuJC6--mXWY&aWl0RV z2zi+o?5gPwaYXm(Mk@{M>a=WJB!3jeW<8xGd9AsdPkjt--anH0rPnzhLc36U{?p=v z;tUgUq?AZIHfH<~8Gm>IyF>j=1k;~CI3t95)kmHlktsvxTQeW}tMM zT7-vtip`7|sY}#OVl)Gjpn|G3Llx3b{^7Ek zNfiChCgop>;Cv-vU9Wo=Q%W)zw1=N=IvT)^5c*?#zlsAO(jC*gASbC*==w|Og7S5<@vK!YJdDNl zu>^}Dqw>|=$7tMx{<@0KU$KZMU7{&?sCO2wSZZjJJG7Iw=OO%aS?PRN7gY+wxy{~U z)q_JDZ)D%)^+2S%|Jz2h0^@$tF>WB6i@#>@v5fhTg#3Pe*d-g$eL=#khx4xu{q*Qm zysoyL4vTF}Dc&lnIzTphaj}rC+ZR`6t5ZR91!QR@b}>g-to3Rev5w3(Dt>qK?7_Il zE89=(iPP#Ow0!I$r=1yeqh9m97WLY})es9FPv4AQ(jO(3+-#jJoDjq&@1?nuqnCY5 z2k%AZ=kxYWorX}oKM){)T5_hM3>|j*78PYuE0iKr4 z9Zo`>2`|u}M<%Yq{;ckL1gZT6T>y`S1AN{jYggk$&Lp3CY=uxtq|U7qu~5CqY{e;^ z+ot2Eh~D=m<3gE$dsCkjYYi&a8mIeZSL=cu#3Bd=*0F(YZTR9-Ygc%X~eiXi-<)>szs%-_rMVBvC zm-l)yF|X7V#RJ_NzuLW*@2{iV@~tNz63FbgXluyz59eA(;E9DW2{FGX$9;VkW;JqA z3=nN6_j+$@b2@&3R=K@3er?qE0HsoMJXKyK>y4vZVk>aXz~rQq3C51$Ky7z-*P&iL6X860wStlGZTT|}hZru*)w;-W3% z3z*ZkNx>Y39*)IN^|#gn`+bWwdaeLPP+4Dy$+dN-l}malJe8FTulaZ$k&+_^oht6D z_Sjsq@0fksoSI{npO8Z>x;R`TDfq+JLKPf@=%J4p(}9-sY=X?^N`z%9dx)F!Hg$}# zk2!?~k+iUI;oLYoFrX3=Io{z;>GEO(+c<~T_{`f6DS@(2txr07xNnMkSgky;ehZy$ za+*2x=fYE?4DQs0^}-Q~^Hnd!IgI+{|GcGI0&Mf3ribv}k&RRyFco oBbQYHlkUGZ2Ffi_@% diff --git a/content/ko/case-studies/openAI/openai_logo.png b/content/ko/case-studies/openAI/openai_logo.png deleted file mode 100644 index a85a81ea063d0cd80218f48c376547313a47fd73..0000000000000000000000000000000000000000 GIT binary patch literal 0 HcmV?d00001 literal 19818 zcmeI3c{r5&`^O(DjV&`0N@`4$gfW(pu`{-V2+39qGuB}+V<)l>Av&cfYf%o#l6?(P z6qQ|;QYeW^h%CQRoto3>ocVr#*Z0qvYs}m}pZEK|pXc>i=DD8hdZG=FXtA^GVgUev zT}K;jMEUNg`~`y;DW8m%m!~OT%8eMl@UjZ*&(DPp|Wz~2pKua!_uNqNeKiJE+Gv^$cZ7OP?Wc{1oYbrro=*t6g;u^ zC?mAyw{n#K6k(1ek~<0x_wn%&_mL7Ocsfw*a&mA9Nw}n>7)2pQ^m8L&e8t>|`@Sam zGY=X^wDWXwCpi(^pc{EHwgfMdA`G@s==%+4QmI-;c>1wHxiNJL;TX`?nodJh>nDRDEYnn9}82)L|^}V?Oz_RtLraI z6G^NVif3R)z ze6@X<3SUP^fpV@Wbx$0IMDR2r5b#Q0XPM!j2%+lg8?#DATm~wvkFj%d+aP;T0sgb? z$C&@fh(lvYIHip_DJCIDd7B`RC`oCQjI5Z1EJ{LxD#=&nPdW4nSSNeGU*wQRpb#?D zIjEFB=b)@1EQW;nkFro}sq$d$Q1%2*R}4wX$ra;(gS)#qD8PR-QYpXBER;F{Pw=E1 z7EVb@0sgCID%Y3UhSG5(k}z&|I32VSr9s@u35&9qLrUA>q$$f>M)I(jG!lyw!ysjF zVhAJ#Cu@KBu&uN$?h8bHeQNvPQlklWUK=asTWaionVQP}x6}rnPL$magQwDL9Q}V= z;umTDR!ZB6NSUL4RCAXygTL%&#yF3^w*J|Hclxp~xnn$uxQ$h(2>Wx{ec#Bw?^7F| zFRK@YvD?_ul`6WtPn@a)WlsF3>-}ryzgmATp#b0b6~H_FbuhlR{&kH0oV9-+Q-zJq4Mmv`lw-qu z83%*{{BLbPDu30RZHz14>x*7iQiAe=BQ`$2S*g2!wwnLDmAd<9>ksEKHz$&k7fnRe0y-{g8Z=`%E}Dp_ z1$120G-$?jTr?3;3+TA0Y0!-6xM(7x7SM4~)1VpCanVFXEuiC~ra?2N9}Yjq889`QPZFq({a&6L@l7>qNYJJrsJZCh+06$MNNZdOvgnN z5w(Dhi<$<_n2w7kB5DC07c~u<@xS6?`SB1S&W-ZqpAY2`zuBzza?0a@P&;iSeE{$m z0)UV(09ab5e7^<&F9`scvH<|pB>>@JWI%riB-@)k~W&A^f?uYZ&SK>@QoVzZlAtI1)4Ro;? zv0oL*6ppiOPQv1c@xwAwDUYoJaz0%!5`&&>|=i8?)L2oSI z?8e+Q-4rqA9rlq^_&k{(pf3-5eJJE+vz>lSDR+r`>{%D~$@hJ3qxgUmICVv$ZBT{r<_@bz;JK zjP3NIgk-e)E>K&pB5etgDIM%}NHTVcDP+-Fz;~0WloG zpPpUi7~vVH4vjBu<`14+DcbJ9BXVrEMg|>uuA2b>nc0EwFth3k#BX~$I^W+qGb$tb z=yY53`Wb@aaa|+8CQ1GCfx6|F2MC`BypE$4f^ieGb;02Y`I-H^00e+-{~ZVgGRV-D zL|npn&milqD}gUQW-I$G%z0pE@!UBb>kk*>t!h1_;zeQ2iCfSB`6@`zFp&WYJEVDO zym_FVaHaBNt3mMc$o=;VTx_ue5p2QEMNeWb^i0W~dS{E($Fjqi0oE9@&xoIQM?hy< zgvRF05J)Ciqf3^jZuRO_&+Ac5vn!waO9z(JxCCH~AmEJeY$~xselpbolaw(BZm`(C z^K;znI@Ygr@OeR0Iv8LGFR-!Y*>Wzu-Y|FAONZ?Ty{mEm3E)zu*RNG28y zU2SNBe|?2XrBeB&oLm|1aOOmCe2w>1d-~d^x7wSm_9JVYwaQNB1|0wK-0#b zr4>0SFGd?7kZ66UEj5eBb0+JTn$oj#51>_461Go@uP>vgk8m?K+$UaC2|cW8!}aV| z@KSX2V^x~=y2dgvaA}`d zUeemrFRR39-ZilTIhqSro@?HR=IW0Cxs+lteqJZu9RR@j)+R!>!?(QU;^sFvgzYM;S*B>BLwZti?n&D& zSYe?A5A3O>2tJ3zu)kGPp#;hj5+Zu67v+eCbX>^dgrv;JIWVbge)m8=>T~(vNq<%k z<3BRv$Di{9f*&3*s%VF5N5**6EFEWRs~wX|h)quQdGkh61ttiFV8l!&i`*?`_@BVB>+C|MYZL%+z$?ObFMQ zu>SBpmC?FretD+lfB@tfM0;LO@=`YIaR*Zo7sne(DByN@ME|ZVMR~3L-fZmiK8{C* zE?J{|{8&`B&Ftah+sT!9NNSEy?~|P**wdVd&~S0?nBBWKwI3lZX&9fi&7ro+wEf8q z5cy-X$2pV>@`Y__61qr31-$=STa}Oq3?`~g?iPX+S&}PEb&0_{?D)h)k0VD%z&oD&|t%EEPzPb=2@;(f} z;(aJzsKA)1RbjWYvs)RVCJH=&1_X+|^T?oH!zMCtBRj5L|GAF}_`UV|=L>{-cj=gX zq2fq22hrk4Hgokvgigv@SMb{O$Gf@n0e<5z2WIPit$;ByTUHS^uBe6A1&_6LVLQ6p zT;;I$*tE~7!2F<6v&~>NE3@3dH%sa3W6c9`8XRsFq=;aj%WX)-Fao3T#EQ4-!M2$>O zL$8UuF5U;oui8FZrJ%s>Y3gIL=-kO}KL#`!Cer-a#~L*ec*f!n6BlKLWzAeLA4~-s z((N~^7z)jhxSlo;09I{K+ntK?8DEZzNa}%s7#U%}?8$1?NHrozirZ%Du$PiV`>rOK z)=BxEQ#hlop-?pdX6yI{yURQ*IY=EC>({l+R>Nd^pYiF^$d%K!XHbQLiykK(+Fa5g zEJ_v@(}y?@w(!14<~TC_X@#JCDQ6E@C1c3=N{Z#DJMFE%cMFgON{n(x%*@=*ypl|{wl+v*^YxM1VwYB%H~wIBSfiR? zWBRFAkF=Bu>x+}148Zw-m;)VFzOO&YqXB4Q=5`0T^3tp46NqhhYRYz?-xae@^xtD) z48!V+V8ExZ)RDnYI;wDaXf{QI`)0zR+o^_eymh>$^~9l?i7+-!FlI)q#X4ZB=te29 zYv;6OUiN6f>mK1^;ZUHfWbPrl1}U$FKK;gNDt-0IfyKD9Ktufgor|S#rG<;5fhR*C z4d4@LDXvv7LO>y0y_a;m&hn$)Rch_Q%SU$xuAI)3o#;H7BH03HPYrqETzxZcJ?HRc zp4)SWm+jiCj&u{SZbie{?vWYFU@PU}=;2|qC=qP)FBcCw~uV;i!tEps+iqf9bCnV<{$)L|q zH!?7cExarNL9=X5Rc8w|Afuu(Jx7@^yjg*1Z2?6OnIRb#`hthVgpe(qUXn}hzJ>5J z@?@QJroUesn88=*lf(FSGeLH_sO`UzU10XAo2R2U35XFGd}4HCD>gwBeyc|-9RJ6W z>jaDCtsOFl>Rpp9&0heU83xN*HCsa%mXXdHWHXm8!7YMlv%Ktw$|_vbf}E#~auW^{ z`t+Z{R0sn0V2EaO3ejmSqx+hy3Xm9)?02;Z%Hp<%kx`(`NG=v1{c>F6=7gL0u^S_q z?H5!|`a`oGz{R#6i!1FxxBo!te>SU{)as5+dagAg_w3&gdV^ZB)n=2mxu8 z5!u~tBGAk*U{lsy9i$Cylzt2e#m5cYt3C5V`_kh<364w1RS;ta0E2RZ2+}$y$@}iL zxd7pU?^9nWhX_b1mVfG7|G>xqvN>_F+u*QC+ekohTxHZe6FXxVDC_=09vVM$KKva^Ke9lm$qsZ{PEiFd3S8Bl2C?On@3A!HR`;D)}0KZtS8lNFZdIhf497Fv+3 zHN6XF>wV&b`Qpvwd%K-+_dh-5DgyICY_~H>Z#@^cu=7-9QdQq5YH!;?m2)kQxs(e= z!TjiF0p+LUS(7irMYtFwD}_VmLkrvhrmCpuXDzq!!ue7MVWPHV6=aUjYGAhIG~ZK2 zg$F&`So|kfkaeN#UA!4#k?rEekM{Ro*4~cGPF6j4ck&&?+?MHD%aJ#C1o=qY?v$D* z=)40gxV0u>$0cI_V99WIsbcd>2Q!2yo;|xw>HWP;Bdm}u?3^N;3I0W)#~tjg*T$VP zuDHMP787VQdX-E086oa2#^`G{bXU7p$i^jwWk|cUUon1;gy5gVH$eL;;|}@pi0?HB zf$mM@gJ>S^^f;$6p3nN^zWTjJOjTq=eRiNz3O7+%#fBp$d@fO8RX5O8Y1A9n%THq#bYzU-cnRc*nD|>#Q=~6={Fe!12<6Hs_jTq_>PRYWvFNo3&N=Ws4MB7fWvT-=y?PB7NBL zABY`Gh4GTyS)?}QrF-Q$6Qkz?geVuzeu)9Nga4bGygN z<%QWQA(%z}na|xzYjtT4%YEWqbkwvj35m~$aSz(K&BU;SdL!5)fhOC^G zc0pqEY+~illqZ$U7Vgfs@qV~oBrPK=p@e#NNnLaQHOht6!*T;7YyICR0(Q!FKKZSd z1yISYB<&Nsa9@i7;M45p zO!JV$v|g8J{#e;2j}nQuj*p@F<-8Zjlrqf^9N9uLPUlI@T?ydkFDf+A;pEZK3{S=t z4n5^0?NVi8_P`VspWY+UV`0n20OBlr8*eV`yH~WQ+h_MKhs5BuTUU7sI(Df)7%~=F zS=y_3LZsO>`WW_c=f^eh$(_rL6@{m?5Xop~!nR065+$p3Q~ibd+Zd(d!LEyJ0-WA>SkB2+}1#Z374 zW;&}zHang34*V^_JSnN~x9kdDt%&~kZAq5)?M>Qsc#}x!9iJ+fgkgH+irdc*kxlccxGCmR$xX-C`a6p;0 zq0G0de=`JiZQPK98RptUs`_js1yl0=R2S?!bKNftt@3O%cv)8+d)b~FeJ$y*G?#g( zRjsAVta8Qm@xEgDJD|-7Qbs zNf7Q-1v>JiqSxAkySew@V_f!edN6hoS&d8P5=)de3YZXnv+5FNfbF4IMJ;+Eb++~U?mu-<|{@`liB^~JrS<#jKy_#45VfiItQ z;|-&FFJ-q)-6!S^%=Mv}9`D*d7?LIwKc@bD%BY9m!Fyn^5uwJWJ|baiB!IW zDB&1=Cz`OlP+9!P+=FSw5|y!`;T@3jxrIA*M$Zo!DEYr%u62N0EBpBUQEZymW5fiW zh>Z@9EHoi&F@2sHu71w#^fJ~p+9 za+NM5e+=UG9+4!6=uEX}GF?{Y=NH*~WoW0U=B(dn!O=mH0WY$)TxX8ftYU>>qI|?4z4=-$c%UwP&N4bK`wjg#$g$cy6 z9<4Nab!dlE*W`mvtD3d9dsksKEg4CI!SAZijt1NKhfBpC|P!$m*eAnbK>x|`PM6+?}ILbg`ctk*xrKNR1xkMY*Kz-AB2n7F=SI zuPpLJ=CzhbU-m?x_)Z2DX{ZQ - Infrastructure for a Growing EdTech Startup -case_study_details: - - Company: Pear Deck - - Location: Iowa City, Iowa - - Industry: Educational Software ---- - -

Challenge

- -

The three-year-old startup provides a web app for teachers to interact with their students in the classroom. The JavaScript app was built on Google's web app development platform Firebase, using Heroku. As the user base steadily grew, so did the development team. "We outgrew Heroku when we started wanting to have multiple services, and the deploying story got pretty horrendous. We were frustrated that we couldn't have the developers quickly stage a version," says CEO Riley Eynon-Lynch. "Tracing and monitoring became basically impossible." On top of that, many of Pear Deck's customers are behind government firewalls and connect through Firebase, not Pear Deck's servers, making troubleshooting even more difficult.

- -

Solution

- -

In 2016, the company began moving their code from Heroku to Docker containers running on Google Kubernetes Engine, orchestrated by Kubernetes and monitored with Prometheus.

- -

Impact

- -

The new cloud native stack immediately improved the development workflow, speeding up deployments. Prometheus gave Pear Deck "a lot of confidence, knowing that people are still logging into the app and using it all the time," says Eynon-Lynch. "The biggest impact is being able to work as a team on the configuration in git in a pull request, and the biggest confidence comes from the solidity of the abstractions and the trust that we have in Kubernetes actually making our yaml files a reality."

- -{{< case-studies/quote author="RILEY EYNON-LYNCH, CEO OF PEAR DECK" >}} -"We didn't even realize how stressed out we were about our lack of insight into what was happening with the app. I'm really excited and have more and more confidence in the actual state of our application for our actual users, and not just what the CPU graphs are saying, because of Prometheus and Kubernetes." -{{< /case-studies/quote >}} - -{{< case-studies/lead >}} -With the speed befitting a startup, Pear Deck delivered its first prototype to customers within three months of incorporating. -{{< /case-studies/lead >}} - -

As a former high school math teacher, CEO Riley Eynon-Lynch felt an urgency to provide a tech solution to classes where instructors struggle to interact with every student in a short amount of time. "Pear Deck is an app that students can use to interact with the teacher all at once," he says. "When the teacher asks a question, instead of just the kid at the front of the room answering again, everybody can answer every single question. It's a huge fundamental shift in the messaging to the students about how much we care about them and how much they are a part of the classroom."

- -

Eynon-Lynch and his partners quickly built a JavaScript web app on Google's web app development platform Firebase, and launched the minimum viable product [MVP] on Heroku "because it was fast and easy," he says. "We made everything as easy as we could."

- -

But once it launched, the user base began growing steadily at a rate of 30 percent a month. "Our Heroku bill was getting totally insane," Eynon-Lynch says. But even more crucially, as the company hired more developers to keep pace, "we outgrew Heroku. We wanted to have multiple services and the deploying story got pretty horrendous. We were frustrated that we couldn't have the developers quickly stage a version. Tracing and monitoring became basically impossible."

- -

On top of that, many of Pear Deck's customers are behind government firewalls and connect through Firebase, not Pear Deck's servers, making troubleshooting even more difficult.

- -

The team began looking around for another solution, and finally decided in early 2016 to start moving the app from Heroku to Docker containers running on Google Kubernetes Engine, orchestrated by Kubernetes and monitored with Prometheus.

- -{{< case-studies/quote image="/images/case-studies/peardeck/banner1.jpg" >}} -"When it became clear that Google Kubernetes Engine was going to have a lot of support from Google and be a fully-managed Kubernetes platform, it seemed very obvious to us that was the way to go," says Eynon-Lynch. -{{< /case-studies/quote >}} - -

They had considered other options like Google's App Engine (which they were already using for one service) and Amazon's Elastic Compute Cloud (EC2), while experimenting with running one small service that wasn't accessible to the Internet in Kubernetes. "When it became clear that Google Kubernetes Engine was going to have a lot of support from Google and be a fully-managed Kubernetes platform, it seemed very obvious to us that was the way to go," says Eynon-Lynch. "We didn't really consider Terraform and the other competitors because the abstractions offered by Kubernetes just jumped off the page to us."

- -

Once the team started porting its Heroku apps into Kubernetes, which was "super easy," he says, the impact was immediate. "Before, to make a new version of the app meant going to Heroku and reconfiguring 10 new services, so basically no one was willing to do it, and we never staged things," he says. "Now we can deploy our exact same configuration in lots of different clusters in 30 seconds. We have a full set up that's always running, and then any of our developers or designers can stage new versions with one command, including their recent changes. We stage all the time now, and everyone stopped talking about how cool it is because it's become invisible how great it is."

- -

Along with Kubernetes came Prometheus. "Until pretty recently we didn't have any kind of visibility into aggregate server metrics or performance," says Eynon-Lynch. The team had tried to use Google Kubernetes Engine's Stackdriver monitoring, but had problems making it work, and considered New Relic. When they started looking at Prometheus in the fall of 2016, "the fit between the abstractions in Prometheus and the way we think about how our system works, was so clear and obvious," he says.

- -

The integration with Kubernetes made set-up easy. Once Helm installed Prometheus, "We started getting a graph of the health of all our Kubernetes nodes and pods immediately. I think we were pretty hooked at that point," Eynon-Lynch says. "Then we got our own custom instrumentation working in 15 minutes, and had an actively updated count of requests that we could do, rate on and get a sense of how many users are connected at a given point. And then it was another hour before we had alarms automatically showing up in our Slack channel. All that was in one afternoon. And it was an afternoon of gasping with delight, basically!"

- -{{< case-studies/quote image="/images/case-studies/peardeck/banner2.jpg" >}} -"We started getting a graph of the health of all our Kubernetes nodes and pods immediately. I think we were pretty hooked at that point," Eynon-Lynch says. "Then we got our own custom instrumentation working in 15 minutes, and had an actively updated count of requests that we could do, rate on and get a sense of how many users are connected at a given point. And then it was another hour before we had alarms automatically showing up in our Slack channel. All that was in one afternoon. And it was an afternoon of gasping with delight, basically!" -{{< /case-studies/quote >}} - -

With Pear Deck's specific challenges—traffic through Firebase as well as government firewalls—Prometheus was a game-changer. "We didn't even realize how stressed out we were about our lack of insight into what was happening with the app," Eynon-Lynch says. Before, when a customer would report that the app wasn't working, the team had to manually investigate the problem without knowing whether customers were affected all over the world, or whether Firebase was down, and where.

- -

To help solve that problem, the team wrote a script that pings Firebase from several different geographical locations, and then reports the responses to Prometheus in a histogram. "A huge impact that Prometheus had on us was just an amazing sigh of relief, of feeling like we knew what was happening," he says. "It took 45 minutes to implement [the Firebase alarm] because we knew that we had this trustworthy metrics platform in Prometheus. We weren't going to have to figure out, 'Where do we send these metrics? How do we aggregate the metrics? How do we understand them?'"

- -

Plus, Prometheus has allowed Pear Deck to build alarms for business goals. One measures the rate of successful app loads and goes off if the day's loads are less than 90 percent of the loads from seven days before. "We run a JavaScript app behind ridiculous firewalls and all kinds of crazy browser extensions messing with it—Chrome will push a feature that breaks some CSS that we're using," Eynon-Lynch says. "So that gives us a lot of confidence, and we at least know that people are still logging into the app and using it all the time."

- -

Now, when a customer complains, and none of the alarms have gone off, the team can feel confident that it's not a widespread problem. "Just to be sure, we can go and double check the graphs and say, 'Yep, there's currently 10,000 people connected to that Firebase node. It's definitely working. Let's investigate your network settings, customer,'" he says. "And we can pass that back off to our support reps instead of the whole development team freaking out that Firebase is down."

- -

Pear Deck is also giving back to the community, building and open-sourcing a metrics aggregator that enables end-user monitoring in Prometheus. "We can measure, for example, the time to interactive-dom on the web clients," he says. "The users all report that to our aggregator, then the aggregator reports to Prometheus. So we can set an alarm for some client side errors."

- -

Most of Pear Deck's services have now been moved onto Kubernetes. And all of the team's new code is going on Kubernetes. "Kubernetes lets us experiment with service configurations and stage them on a staging cluster all at once, and test different scenarios and talk about them as a development team looking at code, not just talking about the steps we would eventually take as humans," says Eynon-Lynch.

- -{{< case-studies/quote >}} -"A huge impact that Prometheus had on us was just an amazing sigh of relief, of feeling like we knew what was happening. It took 45 minutes to implement [the Firebase alarm] because we knew that we had this trustworthy metrics platform in Prometheus...in terms of the cloud, Kubernetes and Prometheus have so much to offer," he says. -{{< /case-studies/quote >}} - -

Looking ahead, the team is planning to explore autoscaling on Kubernetes. With users all over the world but mostly in the United States, there are peaks and valleys in the traffic. One service that's still on App Engine can get as many as 10,000 requests a second during the day but far less at night. "We pay for the same servers at night, so I understand there's autoscaling that we can be taking advantage of," he says. "Implementing it is a big worry, exposing the rest of our Kubernetes cluster to us and maybe messing that up. But it's definitely our intention to move everything over, because now none of the developers want to work on that app anymore because it's such a pain to deploy it."

- -

They're also eager to explore the work that Kubernetes is doing with stateful sets. "Right now all of the services we run in Kubernetes are stateless, and Google basically runs our databases for us and manages backups," Eynon-Lynch says. "But we're interested in building our own web-socket solution that doesn't have to be super stateful but will have maybe an hour's worth of state on it."

- -

That project will also involve Prometheus, for a dark launch of web socket connections. "We don't know how reliable web socket connections behind all these horrible firewalls will be to our servers," he says. "We don't know what work Firebase has done to make them more reliable. So I'm really looking forward to trying to get persistent connections with web sockets to our clients and have optional tools to understand if it's working. That's our next new adventure, into stateful servers."

- -

As for Prometheus, Eynon-Lynch thinks the company has only gotten started. "We haven't instrumented all our important features, especially those that depend on third parties," he says. "We have to wait for those third parties to tell us they're down, which sometimes they don't do for a long time. So I'm really excited and have more and more confidence in the actual state of our application for our actual users, and not just what the CPU graphs are saying, because of Prometheus and Kubernetes."

- -

For a spry startup that's continuing to grow rapidly—and yes, they're hiring!—Pear Deck is notably satisfied with how its infrastructure has evolved in the cloud native ecosystem. "Usually I have some angsty thing where I want to get to the new, better technology," says Eynon-Lynch, "but in terms of the cloud, Kubernetes and Prometheus have so much to offer."

diff --git a/content/ko/case-studies/peardeck/peardeck_featured.png b/content/ko/case-studies/peardeck/peardeck_featured.png deleted file mode 100644 index ce87ee2d47f7b5701843b8153ed648c46444bd9a..0000000000000000000000000000000000000000 GIT binary patch literal 0 HcmV?d00001 literal 11786 zcmch7WmFwavn~V)?h**@ZoxG;o1h!_Y}|F@7Tj$E!QF!sf^8(YyL)gC?sj?4Ip4YW z$NkP)_s^ZRdb+1qJyq3JHPbcS9igHmi;hB!0s{ks4v>=qzSVIt`i$!@$5XSgC7+wG|ZvO&sjlK>yIN zdDuC=pDQ)mYjg_WrYl@^a8hoYmTnWdGSm$R9gmy)`P zmyL;lDV3Ngg|LU<8v{EtFo?p#&eq;V&_jgkKllp1wg2g6r=s`|60nU3)qg6bt*AmF z>ELWe!NbPOYQp)6pMp<-jgv=!pNpG?f{TNbmz_g^os*lDgHw=0K#+rz;=eAcH)+nM z=7K;eng5dY_9Q}O2?jd~va`FpyR*4}Vsmh|VCNJN5MbxvV&~#weWPG?@w5knJXq~r zsQ<-5%FM;Y*~$@YP^%C9D<$Wf3vlB`7blQ8I0Wn`lx7QX*7ueAuk4 zOa=M)KMC;i%1H6?N=b5ZO7lrbiA&4yzjbj-^KTb9|F5W#(+|jUn9|IS(`uDm>{9Atig*E;6x=8+8Ec=@>?Ekp;|8V!e zy53^wpXPrD@7u$FhrgNqTg*GZ1$XIb*xUQX|nqDG3h7gnj>>9{GPR{~rkcQQ`kd zO#dmuqW>fOyUst^1U6s3Xt-vz*4ARbo18P{LhyDwNvA}TK*^zs6=Mb1x;=(sir(`d z4p(>l)lfA+FIk5Br8_DY6dH;s>^Q!ifV9oy63!vGC~Y%!DI9~F|=yE4vz7_ zDqOXw^Atm}+)6xYG7f9IhTAGv-I5|<5bFhQKpZ1>eH2MD?<_W|y2*OKv)g;J*@cha z4o6BVBVnpgJmxa5f(7TJrd!}Vd8h(g$miBgwVrbeNZy)2as4GUf?tHB8-o=D-%G?92iSF^7o*K`gN#C^mI~rb&PTRk` zAIBp!js~%b#2GJmCaW;o*6_i>lK;6Jn$#v&%02}3jm)gH4=DPO(_;qV7~X<1~T`c9C1S&wvq4A0rY#+_Qc z>Eqwk*hnUbhAL;gRmJ`q6dBa{-l68&wMY=nNF^|E)hVgba$*9Ew+eNWH1_eq$RUdr zLjbhvp1U&9;M}cJ7Mjmdb2T7i&{G6&Pn5wl>qf-f3zU{Ag{oVHjURNm{LCKM14eBr zi$hfS_3Nt7RMHj70+WI@e4=p!|Aa{(?&k;Ac%y4K3c?kIL2`LuXr+6bJTKI}OEA(_ z?`9mIRL?Rr8{hf)yQrzTyl<93F6zTD2!8MUAOqEiqYhE!*EY@XS%q(73+~F7{J?3+ zpY40GQE)X!=3~wOf=7cKB1x7r4o67>8?u5bZZGYWhbPqN8rVjQf>MMQI1@{~Guz}i z!@%OXwXSzy2X|lZ_++)MNr4rtKv&YSa@7bQ5#g(1v`E@3Egp>THZzM)ee7cLw_*{R z(ta;fCPlRYzj5axjT56w`jM0reRz_Kn@7&EbHabLjf$_%`Kb_U7_dUG(DwS055T44 z-gj-m_F>|GQJ^W4wGjevNrz@8C4u)kyQnJJ|Mbo&CQmjh9f&=)@fqR+fjIP~N%Arn zYrhB1!$SXDdv#lE1qqX#HfXH-vg^H!MrL2TUQgNd8)$&Lw`~L+mVeQowk)5H#6-?W zl*D}Yv3)ngWTqo852-f^i;cYui?fW5iP%jD-ddp>Y|r0($~bK@@h~wFF-1XGfgO_2 zmEiX}iyEbs9X2)TS3F&>RU@!71kG7Qt9dAlwu?4v^ML_NZ(@v+FmrwW&{2|U5}5zDimbjDRoRb9>s!AgP zhekveA$!U!Dea0C;E3h2*T#KUw3mY#;GoiZ{=Tc4v#+=@rG1%)izz){oYg*EI39`X zo&Ealf@(dvUo|mKStc3ff^HxlW2R-PdtP2=Y^Wfgy!^<1YSnKZJ|d924nv-D_l*HG zDd6*0%p8JPmp~c~9-jBNCqfLR+=8jbS54-&vR(~D$=|(H^eI$n5p049yJgDjg#J>} zULciPmCD{`6NCSLg^>V~RF?j3Fo62_=<<&tLs;~mlrY^g?9>VnsAXv=c`Ah@7$l+! zuu;k)_nm&I%YN2AoAyWWAOX?P%jPO~AT6-)4&ceR{@i{R2gheAqxU-mIP&YuH(#x( zFsM)6sI!V_`OUdNK65m=%-$Y75Svf|By>3J6_mX+qY(92g+)KjFu^Gqp;P349`k}? zp$Ja8=e{#y#gr_I=>2{owBgutD`h#`eV^o9ICV`WN|{yn`^YBLp%6`$A%W4Cwby1f3i)FgiSEc zwIL;Ma_L)*`XYgJ@U%+dNi?HTZK}ckTz9yV!6<_u-e(sVxb~-qqhNqm+$+zP+jhZt zN~7&!`)H}E%|2?JuL-kjcV3N0EN8Jc+VON_yYe`hbF%#G?f6#}^C((}zpb8DIL&Cz z&6%Eao7^gV45L*j%X}+fu#MFo|uc_18#i zntxDg?OP2zrrFOFB(eY;)TSe_^$-W85alnP0Cm@$!X4$aJ`5nY?zl_$K{_2gwyK+SW5f~BAaE6KN(wC{}kay5& zWW6uLR(zk-nemCX`l8vinD7|)Fpr_xrds@Tf`XeByEHbu70fFJHH3nSX!%?43QZNL z+U1(bzY^>|!lz+jA!A{L`G?YA;d+?2*r(t=zqbZ)G3IK>nz)N&Hyp3%K1H<^Es{h= zb5k(#l5svT1pA6N*(u#+lHHM#-I3VCXBt|n6Zg>cbnjel#U4z1D|_#EN@*tYYXNYO z;M0X+#c&`%EE{fd$>xIP!*4k;s3NU)TU4}l``3uhekddiOg#%WbZ8j@M*vzq!bj@e zS*B!&tq2W7zj_|3#C(N7sal|kNd=hW5#=?PGWJ=KYu2B;t&oOvQTJxpHy_1Onz@{*1ykl$nBG3!~pyEu_R(u_hifcE9N#A0~bjD{<6e7&%1 z#2k7xcQrKD*SGhURV<$i)^Q!(yX3SB@h(}@Nls2SbAgm3$kS}zrjL$JIcWMXd-i9G zgh$6pn3pJN*MrNvCVmIBM`#)1{}yQC%%(^f*mf@Z(8;Pey1xq5+CMDsBlKP&(kMTNCG=KT8Lrc(=YmSxGXn+LeQ!S&)ef=v+oLw6_0hXLM-LCH^`U#SI7n!M4LpCXXEy4<<6tGR^+UA(I&=d|7W~ z4zY?8URlJ$4@xRjZ@t;gA3@gAZ%!Nl15YK$%KJ`lwGe|ke%pa*qdHY-6hfULOGEF|} zU#+z`GR%0bZLcU@6K&*R-_jGwvM$wzb)`WHZ|2I>!l3#fxfoqJe8oHGK}k74O!rt_ zBNl~Sxu*3p&G!kzSGvCnTFYNMa|2XUpY;iag)d>iOZ+o)Bt4|Js{(&2=Lud`3NxAN zCBz5I=8w>^0ZMLiH${YowhoqDT4P>bO7hOS#9wSxH4Q=ILvmuZfyRvvcW0M7#q63D z18&VL5mGpo+{C8thnDSTLrEvAbuKMAOkvK=s~NY)>m>urHET`ouK{Q|^W*k*#Ve|S zX7T*z){Sm=BSA0lH=KhPdQb0eh#7e*VQVvo_r)*V*cI=`6QU9GhSRk+r?DvZo5p3^ zT4j-!o>=Ul_t!Ql?u9+88{oP@!Nr9`Q=m@V7nwMjM*G!y#Z)d&Qxg~JIziVy*&LrA zE!UwR`EK{8%+3bct}S(XZ$eI66Wc8O#fz8s5H8;f5|?y7zgJ)LJUWrhM+=>1f}_%$ z47FR<6@9u(#E>0|HAj>qs}lxPs)Gi82PCV}DG5R#@s67oNEQUrH>AQf2knoqf`yVL zN-zH7ul;OP201@vAgn?SDZU+kCKArNfQdK?qPU@B!yGA)VW?trA~k3n!s| z(`8OIiyU(Nym6nAsw$;eHqH88^}5SC+%4u$Rf+N`>kTJp?_OF0ITIVN-}$T-b$?u2 z#ZIq9&Z=~ZZ`bennc2#Zd`_DK=`Feg8ca?eh41tf>|&a{Z%*fZ&v&%i_D21QPCG8$ zL}`cUliEL~XT1H_to&>?Bwla6+_P8}zHhnyb?pinu5L3=l_`0j0FNp9VA-L#Ub#ps z(hU$ptez*7IcgIt0e@zUK&D>TulYJ%giD78ifE;i2-|ykx$Pup(VYp#g*HSPdRVpp zjp>b&MYL)v+RhB5aZOuE{$j?k;LorVq?}SB{4y~@s%7LSos+oJ9Wy$5WZ!xjoA4e3 zU!n}y=!ZBOA;pBAtD81_f=D|D)|k2=AuSz6Q}e6736C4GQxJ zyCx4t`bEHm&WKJ7^(OPlUFTb`6g?yF{Z8;@?QQy(A!8B<4DNwXPlKsvLJ%x0p`64E z!jd*J2qtS`px*5`uf|}1*&<)!1v1d|r;$7!#BS|<2W>U5Lg`n*moOqR&A=jYb6tW^ z3}T@B+_q+CC)^GK!ee3q4sKq2$oYbi?9VCgq4ZdY`~}XihFv4$KI?2_TJ)1QlCgO@jCTz(p`$hH`WOUZgVo@`?*wEcvUcYWuEcPwt@ z+1ox+Ka+5x>ePQBuwju2U;amZR-Iau76Ikr#)*^D@TPToN7ujq>VO(D&GpB5VUm#= z*P-p^A^j~u&!Cg-Hl3MfZ=bmAW*3b*ToCw3&iYECWyo3O1h%)|tY!E3vu5R*tl=tO zq)(8^Wo7INl*pGYSG76NZ8QDQ8J3+r>+R6#=WDIYgh3`8OrPype6+qha-Zjz3uNd5 zrs=mL{*s88UH6Nkg9H=*LJ4}f{v~obu(e~-AMG?T9JI{CeWKR8^s%bV?J+W$`EfWr zH717lcug;tjK?SSZgc3D%|}931DtsQ`j6<>=uOtgX0FO&H~Zp{rZ&qL=Fz$Db$Wl> zAK&qgodW<(r!QeGr%U|cHl=dXNxE8$=p13%`K`;utW=ftmFVS?6&1ImYqg;(!L09S z>;AcCSY-aYX_Vd=B*Du^u+nNf=bqmu<}S9y5@;a!HgmR6o4za9;#6&ZeWhBJtCPK) z>L{*#*=$w_|GIX)B@EuXbZge8%iVU4UbAk(6^!!kcSQ=`fw2ObHzK0jwPNTGa@*A| z#>1MQWMZQ~zqX^?9pKDWx`O(@3r06lpon-MOd8kX`K4u_1ikQ_ zkWIVhEnUxM!90BNo{8o9#APr!DFZ2hHJ{7|1Ee^`o2qw7t=)N171B=@HI!+wjTvWmw>C9!#|y`67C!;{0cV5K6SgzF^cS3UygX@^H#tuaxXwHWi^w$I8X~cVQ?PBuPK9N;Ht0 zHgV>9_GJnxBf`718F2{@x(bEjh)kKqgUJcd6zNzSwxQ>DaJ?Rnm0ZFCmmEKSY%>Uc z=)=WK`g$&bn(cFjb#t1BWE@QiiRV-)O=i~rAVrfPFBv}74?4P8)txGNT+;ZQf?^D2 zVq!a2scTEa(KrV+F7s(=YV7#d``ourZuRZx+PX>M0EXhodR=j=coPF|a!;9-epx8f zEDv-oO$DaxjSkZXq6rPo7q$!E_LAyytsoh|N5&=Pw=^_daBm$;9f`tR+r6$oH!uF2 z`}590T-vwtp5*~;LNrgV6kpotg^^??>ka(1N> zqsyzR`bt5P-BR&AiHpSwOh=!dtDtpczGnerV2u7q5|B8R9sW7k9uR@dqSe)+n8Ca1 zb=vlMeY1lOAR)xcRGccHe``@b!uL+Q zPXDVERF95^3eb?lMntemCvlh6)4k=e=Y|rjO~} zZ}DGAa|1Y=au^;@XP&-}{02BdO1INMQexr^4)7V))yd)~*p5#tWO4J>KgG*bRk@LC z_nC;#cPle+pG`AGA@TxQf;IBlvUJxDo#$1MJ^)+dqzW>Sca5tVhX9HmB8wq*eEV?7 zWNkBNbM3`1cd_QWd}4e0OZ;S%rRa(jg>ev=Ce`4&7i-tJD&Q(dBj8>=9`<)pU@UIU zV8oBdZkxUNW*Sm>b_T+aLAU%Hzsg98L&UtN)pg<$u|_iz&Bh``iTzqjLtl?xg(@j~ zSbO-IUSpw$4w=^2pEg2=O2-b!iQP=O99ZF?eau9Ic8gERW66F$#yCIUCIYJC1+JbU z98~xb)ByC*<{#-u(7LKK;!mZCH*Q2p{Te|ul?$dv*HsWGF%7(L4@9QfW z#PtE`bD4VN;Qi_+b2N+4A+Yh?CPrtBH1!@x`K@8Ps%VTwJM7qL~VlV$ruCje8V}(iWht4no*BJB( zse?RPPqVyxjVa-uS>Aq$y;6}h=VaC7;x#cl`BFbG5eyIFRL2G4QeO*A}SimDdc+K41&rSiV;5fP=Pi2_*-KM63G@_>4JnX%8H z8znTw+i&@qC-qpXfHsX&_$tES+`~-%6-p!PS76mQ%M0#qL^PI;!^q*dVj9|OG0}T; za+i>boQRjN@HdaoDmFaE>H>3DJ-SY@mHqL!(q8ghbjqm{Qv~8}MuHoM*s?>+a`MD^ z#U=o@)c02>-l@?;YU>Lh^t9D2^*cOWGK+d;`oQTnln#}^p~+gw_rv}mf#o`nLGLV` z!DVIaAR-gDFqS+!e4`=6NNh^1tAgI8u7?ddwqZR|tNs36z|HGaE}miJrsR7xsZ+Mq z94kv}VH}*jLJgy=_N;>$0k|WoTuX#c*g6s}HU*bw2|*T~hHJ$!O9ycy+#OF(QODiC zERtyy6-UMGf(dv_2>awJJp^CA<+QeS%&PQwODe6746yM_(bVazBpESq9&Xa`SBR&6 z)bFXpLd2W~T}dAch!LmjS>_JFAe1p|blKrZl6L;Uhb8tv_)-CP(#$X8h;pHQcgXxv z(A*Ry?`sWD=!6Rwv5HrZ3Pum{?Ms28Af@QoFx|4Up)$og77wenV*5n_kbGu0S2F{t zUsFjm(K}D3o6Owl1Nda%V3*gQEX?l*C>UfCVM?TwDqSsCZSI#)bBmwJPg{aWY?nM~ zqepC693YzVrEr9<#Ar+AW?Gil#BB6(!X;3=p4)NhgWt)+-w3O+QU>OJ>Ajdh`63(< z7t=y<87_4&)xp&R;g=Nqug&7$Ov{o&aPumixK;=>Xn;6Si3Q7OW7O)!;};Bq0HS$o zMx10V^#KM4OLy(0VE?3!>!jrsm&cQxQFE6o`eCs+IeNyY7y29`*_~LXrl=afi-4k0U zB_?|u@wu5BB|*_b%Z$m~)Z5_h6BXpZCPGe6LILJgh&AxWVf=StQL(koa8vieXWS46 z`4I#Y(`_uT%yZq<;?WEtDQTI!1e(Dln`%4Cz)Ti2CaQW}vs4Ls^akyG;mT>`&@o^+ z?l9qO*Eqg|{P6JQ-@Bt|{UA;h@f;(TsS&^9TGuPc+h!sf8IV!#C4nj#pz2ni#Pb1C z$q<@#VAYdqOqFP|rD~uhd7%|>U(NcV>1wpmuL^F(*?3X{t6UFOknhvr>igm~y-!=& zWyq#%$r*fhv6i-$QqI=I#p6Cr{omfULk3`X8~A`(E%|wxjL)!`=+W>Plkt*xn#mQm zG%u{dan+_-#%z{AS1;Q15qaCS;~~fH6wPjKAAhzm2I%39q#eU-l~`+FgeU(BvWAFz1z7f*5FUrAw`HmTC{;=d|~Kz684{0O&?s1`3ab0{WY2RGvy2D+%if}DJ@b* z>_=H@*R42I#;%>ZU+*go!@6muQ{8S2cDI9r^ZGGpHa`lP z7j}y$1poT6?zS!xO%YN;;W@d4e*f5bveH&mIn@6nO@uz0Tx)7(B{;ZA%66eu`etm3 z_WRZk|NdKpDlJ$cg~f#p9&J@g9D})ibYTmfxrGr)a!Fv_wYj8i3%UHbB(*zK zld&yX747XVGc97r`yH#&svG0Md>lMd=gnd$(6QU@ykdSCr~)o!Z{@VT4Mxes{L zEBaE5ym`D2EOIa^M0|oZ0@58ag>|kt53CMi$}cW9r+~n-*&+NErxc@xT3Mx(vHq%n zXQk)HqYGX0wzH?){ialt$%|mF0Zsso)yEwb{-<(x2uO-E%{aah)TUdhevEt{L`k%R zRe?0Quwcnh{v&$*m-lTv6GV)2dk27;+q92Gr_{9c@JHl!G4A$)8)NaO`Phok#>0_! zvHN~Fx?Urw3!SfcBB!&mbyz8&CN9w!`9tX~ClHbF>nRz4bg<0o z6FFI}>BdJ`h!&@Ktr7Q#*HwIS!memL{AfdI`t8%tqUmC>)U?SmxY+N>8{eU)L``?s z^Qo+~FV+y#qLRj%K52isKW zwaiLirP)W4)t*6mUj9P46f$wf5C2s7s<(b^)_+$%8fJYl>|7K$g$J7@1g_5>ly25% z1B54&76CaY3tKk{t?(e?-Rs_4qbFJlO)VX<;5zk6V?+)b=1;f@s;XxYg6fJNrj^uh zn=;|yF_QA~C{UyH2Sz?DFbxZY#dmdb)bOj`8vF z`UuN{D?;yg%d_a+lb`1EJ;=0;tQsrV1`7W|iUu|OMtNEd{(&jF1 zz+<*3<%rL2!C0@sr&;u;34x{$6ouUPCe5rXS&Qz)x>>ppOMM_Ie1gI0&sXZh#Xy9t z6qzdA_iE}e8)~=wH8evjBLoOroV#^s>RL0DpLdC|)C)_|YEs+6RS-F?fLbbl>4*nv zZM6yWq;@7bn3(kARYeP@_v61qykz>w@bU5OYjstX$ig&2(4xyau$~{U{&pM=v-9=F ztMbh9bt(6ObDKV7G}v{=wjo&31;%FviC6a3Njsbf`}c&C`xP+*tI6;GcC0TP)h$jK z*9NFwx5bHSirmdqWXfh|Wz{=hGp#svU(fl!ZskCAgx=Fu=5t5IMn7x_Zkd?$)`|iR z=?IW;jJGm7u5`DZn_9dtl(BI;iGBPEj7%}RxPASo~ zey`Vjo_xw_5b?`2=N}DkALb{G*9%m35)^L{$Aqi8DGU3>U#K=0m$i*eSQt4eG*En< zL4bCn@s41d65mzuN{{8psL?OMYULCX_2lFRP5#0C;U7O3jI3aixSVCSQwI}TpO!E#aK8oT2||HQl9j)2AvFKh;TprzEL5|WHgm~sF%*ng5G_#!pW|=Y zH|LY`fwB;Y08}$;GHyloeIF0Y+ZZ0vpuLE~l#osk5mKG9ajo@{0Rq!j=W=%L9W|`H zJE#OaX3(oQEpypvT<;sn<}9O-ZF*@=EZF3@t8h4LNN~X?Gg_&aK&MTuG{vMufh_PO9hp5}r_TEwb^fEV}r5 zus3W{vgYO8d*4hAC2gcSe&l6$I6XooEMDcXGJacqdg*H2V{ea{+W=SkU&ru$0^~E0 zuvzSDPd60W^)8um%jd^TjwacpLR}B2^5lLmNupUi+?z+4J*O~t#2VCEyYkvxf^r50 z9n^4ePjIU~Xk`k4wM*@OJr&wfR*cg6Pt!e?82P1X>g(b}HLL&~CYr<0iS#{JXXkl8 zTs`2Tz!8-5oMt~-fB|xH7votXP))kz?5hUYl;tcj*S_d6rkJ@`t^}DPb&SVZ8j#1m zQMG0($7`~BB`*rx1MQ#0-p0wFns1gwu3krIR)Eday#1M7z)|0V6)nR$5(Cmsmo&W- z@1bR&pf#FU_=8t&W~`A6akY~|g7t?tUvt4TONQmv`N45|ARwJs-Cpb>WVhM4fIg)_ z-2AX`lFyaKa;ac*Y6`=1)o%WvrF41(hPVi=-e;fKAU^{`CbMnu0TQ0b*g@BzgC!R~ z?y)cc?_*EaQX#N`GUS|P#1-0qZUHdfIWzEF4gYx!Z@GU_eLm z@ZDrP7h_Y+%A^TQZI#o}!Ltv;G^8)T|Du;cv&h(%4(-`p2u zovw?9YU?i}-6%+uwAeHY4IT{PjOuvO%n|A_DMl>fzNFIlow9?~$e)UpS$EWW&t>V5 zk9>)5_D90L#A@PEexinTU|)_+F)~2Ost{{21>+JX1cJ&wFm`Qrh*1KT1GSIRBPA5B zf^giAh!NWeP&~1$&?L#ten%-MQ~LYj8sln|R#x<9>ckN;C0~Gie(=BO#fzTz;Qk!n z{+(?AMYuU$uU>8QXP*Klp7m-SVE{#g$ryn6^7+6cYGe<_- z#waJCOmg%0%DD3hc)YSn-BqgUTnG<`c-5=*Xe6w5Qi7f@PzTar(4HqY_Fi7k8baeRLUmZ7`nQwbUSiVjO^0NLS)fx?8pjST zZ4qz9%Tgq}D0lY(O!T?eLo3cU#z~Z!1N$NscCfflB>&5v9Fk1XaFdF= z=5A{5-~r2#UKBL;nw3!#;aRzD-q&1lZ5|_eer-@2GcFx=QGAj;VL)uq?A|STtq@6U z`jT$Ms6GBEx)zhh<}SF5ac~34E;u0{(MBzHQ|{C^<$ lM}_})G5&Z@9WZb(B1%Vpt0E@M|2drvkXDkalK2|X00amlZth?Tq>#4 zJ@}Oy21ZEK4Gb`|1wtuIftJ=FVXBknb}9;Mb73lNZe=!Quq4pRTFwIk)bLQzH1n`E z<2R=g6`>Gv6L3Jx}Q9#%GfR(39CHg*9vegQUiihn&+uhJmq76R&0 zGXIkG+7qU-fWOd*TAt|4|n{ng59QFJANitP9(Jx{SD%NxI`@OsIXestMXJ1%xTf2}p*NzN=FIGMI*0d^BsR95Rdx`z1O}hx)7}E1w9Ifn}e)3*A;rx_2n3BTjd2)4JJKKCMVoBe;%tB3ZTORO1QcO%o?(g!yQh~`UUhc|s z>BIA_6{NPsGj|MG@Sm~cG4Vvq`#p)%yJIurH4N0%jf~LG_n=KK_6u+baC{jm5>j1V zp}Q{xA#YwfTAo7DGx%_mGgAO;MEqlv?ww?llfxo2D9VyUm6+5x~4=wNUM1+u&h*6#JeQqCOsr}zPb13Y&G78~10iFSlaYo=3_CH4oL zg{G1yvA21oocj*2E~(bjez?+Vd$Q62BU+#_ndRRmdt;X|HahQ)KmO4iDqq>RXTH=j zx+I_SumrEzc_D$<%;TCSV&HYH;BjTBP1&Rw zwo5efwvKoh_nJN#8TobQ##eJB51ZEM7-e#u^}c&gp%6w*!{kNpXSun|(I6)GK6%7( zFYa0o;*>Z=xmoJbI=R@tlTql<$_04a&1Qb|t)k-DtnyN-V0sW2BDe!?+AOhI}=7O^mjw^?3W?ed1$%RFivsOt~>&~)m&96%?1i?AD zG#LmtIUkS^4px;Y!d1Qzf2zRTk~D$80R7p@wKV_DU*-vg`uU21#Y{k;vf-lzPcQ~{ zXH;ygLSB{bW*?+6{TC4F849@e^`t|grus;?(;S(m($|}qkfg`-hcb{{x6y6B3pnhv za#j7kygZ}2*#JZnjyigKf7^4p|4wbHMVE5}{hj%XPpj&Az`JCYR)Ou(QqgB5WOba6 zIb8VI@2`Mq0f86%Dxm5{HzO4_H+nb19MwUBQKWzNo4*b$eB!z zqJd!F15l>=p~Gi0I|nZ>fMq7Rjm_sVV7|ITqAJ?|XFQ@M0A#eXZdOSY*=Ohcl^D0u zzNV45$S^%-aDViXUpkxdcZ<4t`uWnx>C<4Bhf+k#ok-j@Vs7ZSdKqQv zKqh)7A()6t)_NX$Z1KnSO<@FkU%Ox zg029m&>{}mR9QzB4VE~YtF6q~#00<*uO5@qju-*PL)G){%Xe5Ts@*s1>}5?~-2Qy2 zjE*Mit#SoS&Lmv^`pT5)%Ih$do7J!zOeZrZiqiQly`?3-;1kMX9ZA`n*3`RO4}E)n z8GSp>cfI0H$=N}fc=h}C7z!+1xFwYHNbx^i*@ zz7bJhw9D*%x4?gz^{TG3O=cNQk);HVb8 z=JQH;Hmtr$*?#05s6H_JsxOtYF5qWw!zKe3D&hq7% zRL~dL3zTfLcnZtMAQ#M6EAIbFbho7O2_7EdSxYC3t^U5ayqt!12Fh0Brpvs6aKFeQ z9NTa#%R?8}=DMOR3Ms(CI@GJ(Jz~(VFGUz@_|(N6 z9^NKVp?LDACS$NcS(lwEV_s;@r?;>{4RvRsr6|cvrS`ecqCuTjJ(2}8rpTS7jK!BI zB_(5*I{z3NzDDOS=viP}AjZ3ER)hV^Y1^EG}9HZJ*OOur05 zm;@hpVN7#!a@l0`S`f4Iy*(OYX|izDAzYY#FF?{#&brtbZ_}@0=J)*%_^=YUcy8o7 z_jLSGh^h=5awZrZF}-sO|AXJx#DmE^Adq*86P;9e$dP2kegAO&NYJ(Iu?xlb-5WUa zKzaB&etyvITIbV44<9MI^|@IR9v*%{?h4=c7(W0agT~a$v z4vTtL!%B{`sXQ3p7mmQ~J31=FD{Ea%m&yvJc;NmAB)(AF9;ELFgV!}&q4mjg(V|aF zE!NXt#v_x<@^0h`8x8#Z{R8i>XKZR`f`$hGlA=0QHB#k>;`K+j*1y*}m5I2E7~U#e zc`%r`?Ln=&y*b|$k6P5~oxwT6WZsey4XCPR<9O@jz4TpnXls%+&T(zJCohuDRCZ_i zJQcN?bET9SG>3>mS=g}(7&!?nMl$N3Be>uAxZdzufpy&*D|`o!t`UysWaVE>mvXsq z>N-*m5~>s@l9M0Y+}+*UBLyl{6%bvYt`{?(9wrFA8@DUhu}+l9ZJf=4_yu7zghebM zt-g6+HsHLzfJdjPhQ56wPj#qWr=F!qqkq8VNJt5j_MnT}=0qq=?O>{KI-T%D$=wp+ zEpM2U2zw(uD_8*i@g zZ}0(P+{RsYdo!(`X|tuGMtAeEzjc(_N78s-YTYkHSKV?_2hZ~8Ha1i;_|r1t1NV>@ zvuFucWJA$VKoeO!LtT?(QIX2yBO8Y!bN* z2eDO=&Hl)Xpia9C63VK|lX|<@sZMlz;*k`Auta^1>TSMv?*`pIvx*mdOiLI)&vBJ8v&X zMjCE!N!t(8&&CVcg*`tD^OL+>HeMVKnD^=pmOiy?825O4nIDf3s*wtQ&buutZ&PW2 zGKf%8HaQjeKi!W$nXK(`gLsCb#*?G#f81bYxl zXBY1ZLMDFw`n4KOc7%t^?Rn27kE+E?IdhpKYPb5_+jQl2r>Cwy4qDc$x6^gdvbByq zDllv|k>J+!PtCZFCTNGTcbU{`o_1M1;^I8pt* zGsCH5GUV$zKlJjBJ9Wl0+t+r_Hoqy#b(x@o%QQmKAuWM`^N4VZF=O!AP@-r~#jx|n zufiWcsA>|i4b|jSF{`kIT)#N-f26aVeT1KHY)m8=C)Q(*TV^atQ=hxqZQ)m z{J_o``rC2cYIdftd@pD0K~WVUXzYw(3VZ~f8$ID##SY8brAchT=%C+K+Xq*xj6H@U zYsRqSMej>?n4~0q7we4+Ra2Rl8rbbkJ!&iFQLg&D{$v_q=4&%phe+emmr!qpsml*; zYdJug-2U06HCbmnAhQ1}i90p*dwIFAVI?$?!_YjUeTUXE;*N7kI$+B;UImi0DL#*jy%jn%*mDt#; z#9#GD7>H_^?vt`$kjZM>fYf?5K8vmgH8Wa^tj4O>6)qpm2%;4?-xceeufD2g&=#a~ zgLbt{ml<9+ZM1GTej98!K-^i~FC!BElQ|b{W4opOx$Y)9p02a^>xj}I)nf+y*pOHY z@Zs8F*0*@p3LN)&G;KVMA66r8MTfqZB1~C{IuF}BCzEKIEL8C$gp@^anEpgu+#Ei+ zTnDrQt+6e(!7KBt{xrT8SFW#LqbV(AG7w}W29E&8_VjS@Tw9n9*m{jw9V{$)ZE>7_Jg!)Cw6>QXx0ajvT89h(P~O6go2#UtA~VZp zK0!eu%LqBVbyVUUs7?MLBnQ_UqTHgBOIUxaR=h-AcQn1Fs_=vhxzt+Qa#0nFB53z^vTR7;s zX@?gTN6M(+`pB3-v@JDUmbX8OtQE31rNCnAm%dH*Bj^XbM+P5&J=NgZN6Dy-lYO#_-~7&f zEQ_K}=R|0O?2$ngXE zzcqx}v8SOZD41Y={XkDY@;n$H+tEo!Yt%xsyvQd|#5v;USf0AkQ6S^IdrDeG)krc% zP)OD_bo3*Rdb=<3Kwcin_q2`e^%H3$=JA8|W4E=FlT+taQBKGQN~3lm>mM>u$}4ns zPdvQU*aw>)u9}7#9B;V1wMkFt5o7J+xJWpzgKB)7y`>HhE%T@e*G9$;t5J`ocorQQ zaZYJop%yGU?HEjt6+M(-w_v$Bx%3fr(@!JP*Jf#JC-Z_0b1o$~SXz3QvMFbpo=Ok- zt~Qag*=$@0Q@l==iECNWFr*IJ&+)D3K`LrLv~PosU{g73U2{Mie@RcF*_v zoNitx2)psg$Md`}7B$tI)#Tst1tA96AD+EOZ>dH` z>7#tb?4_*QPoN8h_T+Bzk4HYgW8>IX+d>oCY7N%S}c#E`HeD$`Yt#);WtF@(Je!W zU1Bsi2+oRi1aZXZ655-}n#yBOv@Y)aN1Uswt%YP#wj_0^mMHI=!Td?a&YIRKM$uWJ zV;d3?s-!NH#01IN&+&?NkqtvGLy8b6cA?AWcCdI3e^iwo+>$Th#Gc)3!`WtzY$sYn zZb$BCUaJX;j@8xGxw(Te=eOn-dbLF8FlCUf3c!=c4fg&FH@=)3S2#T) zc6?k`GJD^2e#<-MNH&_I>HxGUP7XYl2e`XkL)YVu$J;f@1sU|0dcd?rDsa;A5+^_m z(g9)j4X(^|H)Z<8z9O?a5A*CI6uJLP{%7mMEfQx+K6{1`;YDhdR^DgPr=H)iq3G0p z3B;)6Oh#2UcXVj`Re?PkcndvS`0vFi<~3Oq+76*hs;AaJL?w_vJq@T-$iOO z%2L_lUIqc7aLU(#KB1O*Y!$u=V`D(hPl-yio$}Yi?1|oeduCj~VG)E`=XWm+!m*6& za0&!)`snqC@-V)GV#&Fe@TfgD^LejGE3FO`xG75?g!7HhrzHWSqyFk>8KqV1xx)D& zYU%N@`5J4b_11+Pi%SNb93lBj`66=?f~`+(*3aF8leWOrJ13704T{$6pVkh_gcIfD znW+1V`0;PD5u}!0qnG7}Yj+V?VNwxtR#tu?0_hZ`^qKGprRuL6ZIvMa;D_pi;c`V0 z9Ubsi7#4eVZqMD3$5V?td?*HSF%}jbXIiizABDkBX*Z*aQ?Pi6Ycv^i53vLRYKZi( zNTb(L>LyPhd+j+ga@%6f-jVB0OZ}8C2p7A7ukNPhJW>$ekB9l7B)o#ruT{fuAL z&$#j06Wbe!AE!2Zrq!X+S51O+a7Z_xT3M3#fejPB>OG-t?(Q2s=&`Y<({ftcC3=zTrQ=~=XW`)BtTd67 z`ZxlG%R2~&@N!LNrW9)(WAP7yD=YtaBP#2mJspKq1tLaTpct#}a;3WGqI%ZTs*J;O~s3pWQD`=b|A`TZ|qr7i;%l zADMhfsJXl_ttC-w>oS2^$LwS;MMU5I4!)~;Jtd#WM#R67}#HQEDhRvNkgMBSOJM$t)f{bIailukQWKsTTmnshT zFrzrFrbIOcaV`xEh>@Fl+}KPvadSz%h9W7qPP)7frbe_zjR7JGw75OT(JN`Z_G1P$ zB#8LFRW0_mS-4|hVOnkm+F)0KNt)^z#ynN8L={T@vZkN6>n)A(M(((FGJ%~7<=WoI zgW-7vOsIGo&*$ODIl+o5tVcx45_uC}fT^B_7H_3VuzT(wD$@-oK!m#JVb~2~efyv( zB*suY?Gf5e>3M$F&mYX0nAUBUe#RCJC3U9CS85(PMFcMhJQ-jFt)KUjGv7g_<*gY) zuGZGO5z)z$n3&mNkP*~m6sr9uyF>fG3;GJNpkF*sOy*S5eL|k$3+*=HLO_I%bcgo5 zUS6soTD;QMYvMwb%vUQ!z#u*Ex;=AcM!z{+ zc?#q|DZ8^^Hh8`EtGVgud}=4hcA(o!Y6(#(_I_EO*SfK$R$$U)^F0esL^_jWD=Wx1 zlI)`bEnptf$8+2L2CpyiPn9`SB4WE+m#Qh-E;O8ZR$Gqxyub!=^n_~^e2OL&zJDH^ zdc3<3j+FqzoZ({!(SpO(DlT&Z^(FEWQjkel^zCUSYgYH-fi6ctV${pL17_lEYNpuRTb}xekEp0}c>_`9!Z;mPyu5x5 zWWyqUrZbKIfE@hjC1L6|oj)%8R71CXvx`n?6&`bR{QHe@sk-$R^8JHWv}_IfD0BFD z`|6+16OzSw!$Su0gV$50=A)YN`RAWYO_*0~TD9`uV70dH8a6Xx)>_NxMCvr}JHf@; zIIr1s&{KGe8+D0HTq%QJlT2>AS03~%1O)!?%C^f^P{@!{e8d5iaUNV4x|+kMV0l;~ z<^1Bg^6q_;Pct`xsR4}m5mC;D$$>;aQ$S6aHUN7So}9VPEBrEj3yHL_?2af4CRt@hH*^;h4A}86Cxp# zwex7a!QNq6b2wkviOg}v+b^B4_h};*8G4luCi}g^x;D-~zn8svyX)Os7`UnYA$k6= zPm+=3vpVLNEImn5e!bYFV^v0q`QK{YAw@mT08_HAu6@8;x9dT=z`y|h(AJW!=Xq^9 zW(&{FtZOFfc2+E)R+&-x=fdl2i@5m4@2*6N3Vr50S7A|P6ZPdgg3s@sjhLjo?6Am0 z)>8_y-__V$U*5GO8uxM}7;hrjpZ*9{WSCLlH!^xn#cdwc=E+dQkrb%e8hh_ao8Qpj zy~mp+N_g`#(}$fMA+WPw|EYUCS=;+P@!E2gwVZ9i@L03^_#K^Qm$wyC;EZuX{>s`f zo3<{ibguQgXDp)xaJK@ zBdy8RG%gH@33|$T`3vp3`l)gI6V`j`-h)5asf~R6AK#9zVmY@Z7egHoc284IM#qQ8 zZ>HQ6wRM`INNM*9^|SVo9!_<=`&-MrIsBTWJqsNPUB2trVc^X<#Q;DuIJ6B001(CH%8Obx4)MQxZPG5j1_4T zr7h=uW=!Lbt&cKE<<^2js3PJX9RQ-MSa39=ppfuRHRRHG8fC|4GbF8lRKg{S^w_vP zD{p_SQZJ;zc}?+QQYpC7EHqk<_mwWb#+I6lc?M-LJLxQdf-D7j#b%3`KVMvN@bI3t zkyC+7c*Vg#g1m>zl`?-d%VIF#|BaA^5pl!*h7=_{{PoJF`Q%1&FD1pLIWn6t=49a9Xx_CXRfC@@MF(&ku*3E-Rt$F__v(%3!l*uOAPiXmzXPZqWFuXo8YQ#Q2< z4?gP`UahULux|;@WE$1?Vo#&9GIezbb3T?^ls|r~j4Xf4aSW%WHh$yvfri>GMob~Z o&#xlp@A5xet!3VdzCnP&ExZGbC_?1_`hQ1WT1Bc%!X)7T070y!VE_OC diff --git a/content/ko/case-studies/pearson/index.html b/content/ko/case-studies/pearson/index.html deleted file mode 100644 index 501bcea8e7..0000000000 --- a/content/ko/case-studies/pearson/index.html +++ /dev/null @@ -1,83 +0,0 @@ ---- -title: Pearson Case Study -linkTitle: Pearson -case_study_styles: true -cid: caseStudies -featured: false -quote: > - We're already seeing tremendous benefits with Kubernetes—improved engineering productivity, faster delivery of applications and a simplified infrastructure. But this is just the beginning. Kubernetes will help transform the way that educational content is delivered online. - -new_case_study_styles: true -heading_background: /images/case-studies/pearson/banner1.jpg -heading_title_logo: /images/pearson_logo.png -subheading: > - Reinventing the World's Largest Education Company With Kubernetes -case_study_details: - - Company: Pearson - - Location: Global - - Industry: Education ---- - -

Challenge

- -

A global education company serving 75 million learners, Pearson set a goal to more than double that number, to 200 million, by 2025. A key part of this growth is in digital learning experiences, and Pearson was having difficulty in scaling and adapting to its growing online audience. They needed an infrastructure platform that would be able to scale quickly and deliver products to market faster.

- -

Solution

- -

"To transform our infrastructure, we had to think beyond simply enabling automated provisioning," says Chris Jackson, Director for Cloud Platforms & SRE at Pearson. "We realized we had to build a platform that would allow Pearson developers to build, manage and deploy applications in a completely different way." The team chose Docker container technology and Kubernetes orchestration "because of its flexibility, ease of management and the way it would improve our engineers' productivity."

- -

Impact

- -

With the platform, there has been substantial improvements in productivity and speed of delivery. "In some cases, we've gone from nine months to provision physical assets in a data center to just a few minutes to provision and get a new idea in front of a customer," says John Shirley, Lead Site Reliability Engineer for the Cloud Platform Team. Jackson estimates they've achieved 15-20% developer productivity savings. Before, outages were an issue during their busiest time of year, the back-to-school period. Now, there's high confidence in their ability to meet aggressive customer SLAs.

- -{{< case-studies/quote author="Chris Jackson, Director for Cloud Platforms & SRE at Pearson" >}} -"We're already seeing tremendous benefits with Kubernetes—improved engineering productivity, faster delivery of applications and a simplified infrastructure. But this is just the beginning. Kubernetes will help transform the way that educational content is delivered online." -{{< /case-studies/quote >}} - -

In 2015, Pearson was already serving 75 million learners as the world's largest education company, offering curriculum and assessment tools for Pre-K through college and beyond. Understanding that innovating the digital education experience was the key to the future of all forms of education, the company set out to increase its reach to 200 million people by 2025.

- -

That goal would require a transformation of its existing infrastructure, which was in data centers. In some cases, it took nine months to provision physical assets. In order to adapt to the demands of its growing online audience, Pearson needed an infrastructure platform that would be able to scale quickly and deliver business-critical products to market faster. "We had to think beyond simply enabling automated provisioning," says Chris Jackson, Director for Cloud Platforms & SRE at Pearson. "We realized we had to build a platform that would allow Pearson developers to build, manage and deploy applications in a completely different way."

- -

With 400 development groups and diverse brands with varying business and technical needs, Pearson embraced Docker container technology so that each brand could experiment with building new types of content using their preferred technologies, and then deliver it using containers. Jackson chose Kubernetes orchestration "because of its flexibility, ease of management and the way it would improve our engineers' productivity," he says.

- -

The team adopted Kubernetes when it was still version 1.2 and are still going strong now on 1.7; they use Terraform and Ansible to deploy it on to basic AWS primitives. "We were trying to understand how we can create value for Pearson from this technology," says Ben Somogyi, Principal Architect for the Cloud Platforms. "It turned out that Kubernetes' benefits are huge. We're trying to help our applications development teams that use our platform go faster, so we filled that gap with a CI/CD pipeline that builds their images for them, standardizes them, patches everything up, allows them to deploy their different environments onto the cluster, and obfuscating the details of how difficult the work underneath the covers is."

- -{{< case-studies/quote - image="/images/case-studies/pearson/banner3.jpg" - author="Chris Jackson, Director for Cloud Platforms & SRE at Pearson" ->}} -"Your internal customers need to feel like they are choosing the very best option for them. We are experiencing this first hand in the growth of adoption. We are seeing triple-digit, year-on-year growth of the service." -{{< /case-studies/quote >}} - -

That work resulted in two tools for building and deploying applications in the cluster that Pearson has open sourced. "We're an education company, so we want to share what we can," says Somogyi.

- -

Now that development teams no longer have to worry about infrastructure, there have been substantial improvements in productivity and speed of delivery. "In some cases, we've gone from nine months to provision physical assets in a data center to just a few minutes to provision and to get a new idea in front of a customer," says John Shirley, Lead Site Reliability Engineer for the Cloud Platform Team.

- -

According to Jackson, the Cloud Platforms team can "provision a new proof-of-concept environment for a development team in minutes, and then they can take that to production as quickly as they are able to. This is the value proposition of all major technology services, and we had to compete like one to become our developers' preferred choice. Just because you work for the same company, you do not have the right to force people into a mediocre service. Your internal customers need to feel like they are choosing the very best option for them. We are experiencing this first hand in the growth of adoption. We are seeing triple-digit, year-on-year growth of the service."

- -

Jackson estimates they've achieved a 15-20% boost in productivity for developer teams who adopt the platform. They also see a reduction in the number of customer-impacting incidents. Plus, says Jackson, "Teams who were previously limited to 1-2 releases per academic year can now ship code multiple times per day!"

- -{{< case-studies/quote - image="/images/case-studies/pearson/banner4.jpg" - author="Chris Jackson, Director for Cloud Platforms & SRE at Pearson" ->}} -"Teams who were previously limited to 1-2 releases per academic year can now ship code multiple times per day!" -{{< /case-studies/quote >}} - -

Availability has also been positively impacted. The back-to-school period is the company's busiest time of year, and "you have to keep applications up," says Somogyi. Before, this was a pain point for the legacy infrastructure. Now, for the applications that have been migrated to the Kubernetes platform, "We have 100% uptime. We're not worried about 9s. There aren't any. It's 100%, which is pretty astonishing for us, compared to some of the existing platforms that have legacy challenges," says Shirley.

- -

"You can't even begin to put a price on how much that saves the company," Jackson explains. "A reduction in the number of support cases takes load out of our operations. The customer sentiment of having a reliable product drives customer retention and growth. It frees us to think about investing more into our digital transformation and taking a better quality of education to a global scale."

- -

The platform itself is also being broken down, "so we can quickly release smaller pieces of the platform, like upgrading our Kubernetes or all the different modules that make up our platform," says Somogyi. "One of the big focuses in 2018 is this scheme of delivery to update the platform itself."

- -

Guided by Pearson's overarching goal of getting to 200 million users, the team has run internal tests of the platform's scalability. "We had a challenge: 28 million requests within a 10 minute period," says Shirley. "And we demonstrated that we can hit that, with an acceptable latency. We saw that we could actually get that pretty readily, and we scaled up in just a few seconds, using open source tools entirely. Shout out to Locustfor that one. So that's amazing."

- -{{< case-studies/quote author="Benjamin Somogyi, Principal Systems Architect at Pearson" >}} -"We have 100% uptime. We're not worried about 9s. There aren't any. It's 100%, which is pretty astonishing for us, compared to some of the existing platforms that have legacy challenges. You can't even begin to put a price on how much that saves the company." -{{< /case-studies/quote >}} - -

In just two years, "We're already seeing tremendous benefits with Kubernetes—improved engineering productivity, faster delivery of applications and a simplified infrastructure," says Jackson. "But this is just the beginning. Kubernetes will help transform the way that educational content is delivered online."

- -

So far, about 15 production products are running on the new platform, including Pearson's new flagship digital education service, the Global Learning Platform. The Cloud Platform team continues to prepare, onboard and support customers that are a good fit for the platform. Some existing products will be refactored into 12-factor apps, while others are being developed so that they can live on the platform from the get-go. "There are challenges with bringing in new customers of course, because we have to help them to see a different way of developing, a different way of building," says Shirley.

- -

But, he adds, "It is our corporate motto: Always Learning. We encourage those teams that haven't started a cloud native journey, to see the future of technology, to learn, to explore. It will pique your interest. Keep learning."

diff --git a/content/ko/case-studies/pearson/pearson_featured.png b/content/ko/case-studies/pearson/pearson_featured.png deleted file mode 100644 index 6f8ffec49e6efbba25af84b21319bcdac403a2d0..0000000000000000000000000000000000000000 GIT binary patch literal 0 HcmV?d00001 literal 7784 zcmb_h1yodDx2H=%M5V!DKm>^yYUm!28j$X0Xc(BGI|Zebl#W3WB&DT91f{#XLmKIN z%Sl{o8f+xpB^`7pl)mNoYy1u&_uK6`&fJ`vm3;B*MX5 zaSYS9F*h1FIXyQ`gteQeITDT~ZH2Ia0~DRiZQvSkb1QGxPPiBr7B13GOV3RYrXpgA zaN;)qBg5_Gkao}D<3lLyuB@WaTfPrByvT$2F1s^0_ z(??ay(#OG4*a|2q0TA;N!68H9)2)DP?(zs0_GRs;RNu4c?3aV0T39%1?CX}3yJXZ0RB3F7;8u? zYY`2o{9m>(J#nC|o12RW2;}MM$?eIIfc!Y(8L111GFE1BHf(zyC>}Kx8<&0wf zTLTo1vP9asxY;3`0e>`_TOi!s#DN%3{~3al3k>!z#m=a|5`{?_$jjUX#KR2+IXV4_ z>mSl6Hx2lIy76zNQCi+EaF7NZg>Xk&V#dRo`EM{LcmKPgKY|! zfLqBU5l(=A^F9YHFuQ ziZsKO!($(s{&v!Q)YJy#>>f&_TlSp4z2{2P*OQfc)HB_e8vbE;ljz}=cT{W&RyZK< z1rd-$8k$QEeHaoc!xYJecRyT)TpdHi@GwLS4@3OJBUk^!`#%HztFoffcm1oz0`i*j z^`E5*VeD^Le~yGd+iB9MPh|;ZklU4s&DJAdVR77HIec{V#E6n?=r+VQa5cLYKnyHV zsQ?LI7JUDWllppqhKRQ0kcy=h{$O9j0kFZp^?ZxbeBUcWJ-3;!fxvy=Wrc@E$biV= z4!~=FZdqVlgMycY2>ptDpNJCu`kT)w6LCW7z!vd?%@v@@1fQb@VW1k*y&rd_{gh@o z+2Vv`FV7uKqvkR(HOkh&Qv>MOFez!VA7s4)^hD;U=I*7&BVFtFJVHHRguF|W|d zpf8ttC3u?JT&~E!%O^h3jqt09Sn5!OzU!pi;f(>OM$Qav@HFY%+YFP91VGI02((s+ z5@V;39}rj8S3qAD6Vuow-P*>Sd^t2#J&3+PJAUt12s#olrCuSw*I>)H$i@O_t5(i^ zGNox>_z;iHikFz67l&$;kr#f}XGU&B8IEj|xOVNF`k2cnO0XmajI6FS!E};=rxoVS2zT4r zM^pS6uIW9O36%k`F0prsFgAg(sEJtjFzPB9x%CR|iU*tg^f0^%jixc#BSV-NRs?R8w5u^PF=2gzE!KY4M;ObCc3d! zp0$;Py-}`LAL_DLRTl2c8^h_QxL>f*N|8h`L=(KCV}TFag5XO%Rd*L$eaVQ`%`NaJ z-;KOr2CwStvl~L;bc&{ zuX1gh6|H5Sst)w*xO|K5*gx}wu7qO`LgTL$hm-_iVL+=_guT|Cd~_vc}o~YiO>v&&fBE5Jkx^X1^oH1UQo&ju|;lH%M9SZ>9`Z zyOvQeTekM*To9g)vsdS@`)9<(-l=pTp7s&p$)pod1xiX@XR^zq2Zx83Q*c|V4^N*f z-QZMzo<*s%U|CHo$c+Ee?E{`l`3ccJR|Oq%LRPG=^i z`4Al&I;$t>Bn~=xXEF7>F`dWLZ@aU`I5O-_e_$6R{-oP+am0F9xvO_;Zwo}(jZ_Z zD@u&j($t7VAatFxu&_Spf#5qbH9FpBBLzNjg5R!L^yPRJOab_%k3eaQw&{oufj3Ho z5l>4cC-hY!y+>U2bgpf90{fpq(iE*j-IZ7qh@EQ}6Ob-0`v(VH?CdAU$59PF+9f&% z=QviY2vQ5i9Q4rC5H2LHO~mI|0$%l%3YU8%@4 zO@y@g1KM(#gPDmOl%n*ABs{JR6!+s3sJcK~p4GZ?(rf@FByd_|@SZmnCdT-97nt6Z zs05b|U^H^k)zy8KU-VQdT!xKe(w|DmOFBSe#=uQvcDOqo0ivt?jn|wWEuS#ivp7O~ zKS}BRs4YE9{FhAcEP!>WB!0J*recjy@77dq?+_soey>%%{GMJq^#ko6g+Ja6+Kp-U za*B{PM?(UVT7w`OF{N@+VSHv4di%4s{9tN<_jy-WjT35;FXcerm-@4qUNTf%WxM%CYB6IP*J6}>V_-)2~P39zk!C+=) zf5q+3*tn+nAvS&Ql25>o@`&E*HCmp6oSn5N-wm2Y8eiB21eB(KYLH=Sab5lrI@fm$B z$!&^*$a(RsTS=Mhu_*h)*LMlpVx0^k8RrKnM36c-Lf4x@b z^`55J+CMW~m{uWd^Ta}jISFv7Z0{Jaff$mplxMiGtD;Ob!nVvA)?9L#Fp8wio+xkf z_2>@>y!hU!dBPxOQSsJ{7QNo8#gipUW04DPhrbpPU@}8eN`NZ39Dv(n5d%>mUg!uRx$9Xx2R}MGDok}8BWxWD}{9E^NEO$ov z_t7<_@Mt`F>KC-2-&Zv?RIl{^srxzPiQcep| zDL%h9g#9mkZ_lb&pQ7B5X~MpxX!L_dvb&B9p2tcJU=7A2}+79VX_w5#J6lHhLEpYMVtt&H^AoNC`vd6i;`%E7<^4fEy z_A+Z(SEdJbBVHA#Dl01=9v+%)M@7AEzBv(%n$I{6NGyvuipb`yu)EvXEQqR!h*<9z z^_)|wGoSSOWxbRn$?d*5MM}R^#r~jVx((bi0?A8ix!u%!ip80Z#zxU|Q_-9LZq#3B zHQklOg7l?{f|k{Nwn7s>>Q~kq?JjR`ClvJer#?1b|1uPtDN>+OWQEl44Huv`*_|cL zi-dFN4DJZiPwcL7&K(YS$PxxrpDZWihAh%cVji1xfoX>)Ra>*I3eN+}h1OjmIesT2 zoh|s0=OkPQ5Bo<_dpo8|Y3fGOxAdSwM;Y-0eNL=$czOtv$JzEz3wY8ls*w&CLyQYO z&VIPlnCFExKQ&Y1bkm-$LbkbgVnT$7 zlYT(cNK43X<>S;kG~&cMfyM;uz(d3gvc9&~6Q_UHmD8HcmB8BK|9eIMEUjSy_w3-? z-a5Tpqvq3CTq}1ZC(pLmrM1OzvNlIDwpvzJ7K5mJjV}X$g@}yw=;YwqMV-HHVXH3h z^QWrkn|3HF=)K4H?J9M>&XgBEGg-5-v)8=1%#iNU{OBQf;m!ee>-+y5}x;gs7jDm46{HKY2k7@!hocD|Xqv8xM>9H|7pw0g_G^zBN>t ztza2@>(4L+hY+*y&wJh1#kE0!gqH>kF|7xDe0;3Y{^H_0xrH7AB#s_E-|lghF=@{? zp$KP`#U$n@+i%*jA551R-U}igN<6aZAL&CZn7~!D?`wO4T;U2>p1el5~4GYuSULe1FB})g zaLpyWx!X3P^-~J-#>(DJlU02n?9($R-tet$Is}F(n1>B34+|qC{MXmKN@F0-4pMI zn^t--_=5PrLdbZA_`g-hLdCz0#bZJZySm4p;KH2#gCs)S-7$?vq zNs%iV!;1W#Q1wbX zwrD3;HurDrXCqamdst8L2wqCtD7=Sb{x4!6)~E?^n8jP6oy9im;4_<xAV|IHd#~D>!Ox;X!g#@JH?%Qnoffl7Dz+DN ze#h-y6_)ZQEsV9hmoW#m@dS;B_U`8N#bixSSXqo_XINFzK6_7>;zq<>*LfSJFAR(j zvYWXrmtLuxaEBATVef|X0#r&gr1SOWT}SMbCLKdIAS|3c7E0MhmTB#Y%~${eA(F^~ z$VUo81Ww`~E8`;7 zN}ypzxinN%ejMhwOM9u|l16nLHatV`Uq|x_7mnbVXoU|2tJPZy1^7Q4`l!~XFP4vb zFmGBg{0)ZSr)Fyzt}?V8*%(g&q1U4B93I6d?7Hm^l{6Wh?D`;NK{}AuH6ZxH! z-Pv)Il8DzG3qUdB=JV;dp}MzqZf6sPB7Va*jgm(*shWZaR)c4n5)#egKF7`OeD^8X z!&OX7$ur(zC1(++*53x{ZgqgI6% z;h$?mPKUr5Qu$)wPugeh%(uU-mDBEA*gEqIP@D_B|GHG)HUgcc>_q%BHWm z+~V8Tye`iSVozo-9=|5J!(=1kz4UJOB(Ab@_sLVZ4cz(6Re3f^I;nJW#H9-MCJ>ig z-|ZsZJV0ymy*`gcrU;NK2!+Z|Zhd{8$oxUD#M0%F(kl8wj$hoAfGkl7Kg^bVRR zJNbj>{LR9`*Po=`)qw#frLa(A(HhXX*TLK60?r?fbb}_`pInegiNM{S27D3loy%Zqn))M!%Q-^VvoDkN~DO=$PI z0fbn5iFk2uSs)79q;5PNnN^@O4hcM`yqiB*XKuSLPxoygS^0em_tj13^Qi9@gzq&9 z$39A$Jj5LoUx|+4PQq^t5O=ehh&1U-08~zlB<+P{8k>xtMi=)167U{&u`! zXW^*8{!evFzw4``Lu6|Pl#H&?sas8I>)Xb;vgG-G1k8}iuYA8X^dVVFgXd{!7~MhK z&~TB3K+;xUiQ&EpuwrBDaS9V-^)ydT5$|5EyDeQ>Muyfb!C`#u)0n-9dcFz7uaoPG zdlL<}Z#r!qsTi-=#G8CXXS5u~{hBPNlQXAl`{HiCeyK^y%zQs)oAcDgb!dojYK3E9 zVXOp7kN^i$AE8jKOA?nWN_WJ>-V46Xd*u79*D`@wwf>87&iWubE<^qHS>xwXs=X{& zS3VsEd(KsEECugWPsaH`4Sxr>X-gt8NZpwrCf%^!3A zmm#f|C@+1QO|#0BG6N^^O`Oi-z-D9A&6U1H9N`DD5W9go;sjRJGz($r@(|N96MWaA z+S+^jxE`55)@7-kgsazyfg8P{4jfRcPi*fUxGL|jslSJErC)BdV1(M z{*5Ywau|sW zB<sLrQOn$U*r(05c8x~Bky)V&UB!z| z#_dAZL>7F%Q5q*zTsCav4ESodgn>aaNt7Xdh1b9_Igj;)>;+l!Y?)K#Rn1dlMWz0N z5J?ZzSo+|zZ0W%b;_wgd4*f8+@9z^-d#((5wh3$e6+6@QFse3jM2s=Ppnk$d2Bx8w zt;d*4a_W9<;ZhJ>BCdB9!9o6=@~sAKol;j`95D&PN2yah>3Iy>mblx~J-4 z=_~{4M4&UyVlC!Kwvi+9wcP*rt(pJ6m-DZEo@3A9TVgO40rQG>>#xGmue0 zoka|8*}aLV&EH2$T;0$~?RyyHo-+KUQDc0znwE85?%ac)(+4GV}=^vYJ3?q|wtDsO^D7GC1_6%ful4ja) z9kA1OzIgb)I262;8)W87gby!STAx4Ll*Xj|I3m85y6!j2ot2V64u$)4FWFsLZqjLS zC;o%eDx5xwo3nK_Q{hvWXgz=^m=mbW+Wfe)cIpp)V&SUh+LT}#scFL%yrsGpE=3G! zP)*}q@Ff+_ep46uLfj?Y( z3G25y?2U4IiEZ!t!=BsYRVifqB1==_HeM}&JD|l2`*#<$(J==l_I-i29C7#iwWJDI z?xf7D$*{&w+Zi3+(jqrGIqA7I60$enbhukRJXT$B>qN~~WLXAtLWRN+3M=QP=b}@$ zIksjd)r`DX4Okx$qqz?hjYEf)0U2?E=ky2B=!Js@N2Vi5*F6{-(W3ex<%A5k?J46j z*U(;R-E&4j21gl~2;2KkRaljnw}9VvbSS#JyMK&{SuYR1;!8_QJG;E3XJyrNb#)Cb zQ-H{8suJkzx2wXT1(+LR@*&42s1>oSgImT{g~ z@KE{%&Rz^WH$Tq=4T2zBj#H}UW@f-}INa3CYhnlFTmlQqVH~xiU!R>dgc5!BI z$|MFw@r8Gh;ZXEGB+mBIHxRGz^D8Iu`#Lb;lamm&e@T0wr!SqEz_10_5{`U~l<_mnWufM;D=hmy;-CagGf3ewm zq~t~Rz`(#-lwhkRu>7;Lv!(5AwR)HNRAFl}VPRqMIxTf|>&}mNZ;_MJ-@6wU7k4`% zBH|tcL$%!~I|Kr$sH%Dm+|4T};JbWdYkSzsowc;N+4G4_$<^JxV{tL-O@E58+lq25 zv-InVK;Tip@|D%qAK$cde;*xj^6-3I??)%1(FOH{(OeZuT3RkCDMlz1dU3R^huQs@?^6>I51B>+BHQa}Mevxan_A_4Nvl!*K zEh((uVZOOEQ18heq+uUEjORS3dj0zK)=X`26@;5xg*VHfc>Tw1Xy@FVDeo+xA(NQh zr*_AyrO?}Fe-4M4Bwd(+*z`}v?Zl`b$!B=&OvcB@x4$JJRfWNHCXe^#(@99lczXP9 zu1?i7HMt%?4lgySr(#`VFk80x<{%Ed(A8f-f7e*wH2*B#51z{pVom zH<(3hU|LmGr0hfLea63(C7Jz`nqz=DE;eSLktSI0PB z;SqZ3PXG$_^pXzNNrWuFg8t+OvX6R)>&CwD=91AdNn$|yt!%X zx~PeH@}3H&t{xE=NB*Q^{HrS>H`R0?p=Or4^Y(-=%hz98TAHjG z4>K|{>g?>S0PsnMp1&h0DfuFfwRFx%M^~4Uii+du(`c9ZMq%3_=6sE`oLgej(niy` z-rn9NVZK@?znjH+ho0`$A6x*uf89|@jCLN!UT!uGF;pR}&P-!JFezjRXHL`M zmGu~M2UJ;c43^1khVf=nynA@Ke2B6o8F)Q(2v-a*_b0-WH48Y{*h0R2~tlK+041c!>U_<_Fi8LWGF(N98LrhF>+GVvn=53IQ(lOgxd{OK<)6Re&bDDB( zMS3-x%l{*6Zpsjv{Sanx2fhDwG^F59y5ELHUTfBURO)NsqJU6BPF?j6Q?(b zKg&Np_yGyv82vaoIr;A0yV*6B&)RSmBCg2*Rf|r2)sLm4WA1{%eZ$ZjQ)|DVAo=Id zpZ7)IdCRMtFXS{O1WMsILAaW)rHR@~d2Zb?FfdSl@#2Z7RFP=yG{dnK-Lk`M$NNpp z!<)>LW`{$HVFY8=8w-oJ?Xj{2VyvKm00N*~iE%B=$a`yq4d95Q228wrXegFN)_bJf z0tzxHHGV#vDVGKfmIvm7Rl(qMDj;W>p+Onqj*~pA@bLmIcx+|Rl|N9t9dHZOb2~d8 zfN5(MHy7xtpq^Nk&hBpf@lSoiGpwiD4V#YA!NCEed|-n){1Q+iRT9bC+S=-0tv)a~ zcz(HfGXxAN1MQ*YX_ClHbJa47wjp3T2uV+WNKVT<5}@;?rbe&a={?zZFB!~2-bzi) zd+4)bu0+3q;OXq~W}iGaV(25n!sq*J3ipyz+}6fkZV;f)he)sQCuQZ+vM8* z0q7Lhf*J1J>jawrj@JS(KuEaE{hXg`A8`#A8E)OC$A*WG?H{QhO*c+B_@t z_X`LR1v~@de3zES6nrtBWmsiH`S;&{M>9(&01#$oW(x=M$2+-1sw|(-I!HBf%Y>npLA|un$ z(*ryt(=E_^GgD*l@7&eUAO#G+1(M2EE-pilu78u8qmK2ztO#89@s)3(*z|_f<|?d| zTCPhc5F*Qs7yV|9rGNId8A#vy1VRkOHl|g~sB4V*>ye8F=B~O~1VFp?I$Y?JCr=b} zM|pZCw>@DKxc2t!?m#7L7z_r~Q&(RftagY&*Ggphuhgo2PFc8mr63SSw?qQ3 z654L5i%};21AnVx!sXZatHnw@x3O`1Fo$un6vc1y@H9J%;YWbC-LF-0CE8_6g3ti*AHbhg949m?;Pj{kypW=ngR8^oS=t^!z zVp%(U#CB}CC@CrB2QgTxJ$sitJq-WaYN%)U*wX!`%rHHD_oWGhjWxU5r_pH5Cb(RQs+FiRB_ZUeG53o1tLpK-)X*qdL%10MT4Q z*oPFfh|4>({PuF}0%^fzYW=GUrYXjc4BxAN> - We are in the position to run things at scale, in a public cloud environment, and test things out in way that a lot of people might not be able to do. - -new_case_study_styles: true -heading_background: /images/case-studies/pinterest/banner1.jpg -heading_title_logo: /images/pinterest_logo.png -subheading: > - Pinning Its Past, Present, and Future on Cloud Native -case_study_details: - - Company: Pinterest - - Location: San Francisco, California - - Industry: Web and Mobile App ---- - -

Challenge

- -

After eight years in existence, Pinterest had grown into 1,000 microservices and multiple layers of infrastructure and diverse set-up tools and platforms. In 2016 the company launched a roadmap towards a new compute platform, led by the vision of creating the fastest path from an idea to production, without making engineers worry about the underlying infrastructure.

- -

Solution

- -

The first phase involved moving services to Docker containers. Once these services went into production in early 2017, the team began looking at orchestration to help create efficiencies and manage them in a decentralized way. After an evaluation of various solutions, Pinterest went with Kubernetes.

- -

Impact

- -

"By moving to Kubernetes the team was able to build on-demand scaling and new failover policies, in addition to simplifying the overall deployment and management of a complicated piece of infrastructure such as Jenkins," says Micheal Benedict, Product Manager for the Cloud and the Data Infrastructure Group at Pinterest. "We not only saw reduced build times but also huge efficiency wins. For instance, the team reclaimed over 80 percent of capacity during non-peak hours. As a result, the Jenkins Kubernetes cluster now uses 30 percent less instance-hours per-day when compared to the previous static cluster."

- -{{< case-studies/quote author="Micheal Benedict, Product Manager for the Cloud and the Data Infrastructure Group at Pinterest" >}} - -
-"So far it's been good, especially the elasticity around how we can configure our Jenkins workloads on that Kubernetes shared cluster. That is the win we were pushing for." -{{< /case-studies/quote >}} - -{{< case-studies/lead >}} -Pinterest was born on the cloud—running on AWS since day one in 2010—but even cloud native companies can experience some growing pains. -{{< /case-studies/lead >}} - -

Since its launch, Pinterest has become a household name, with more than 200 million active monthly users and 100 billion objects saved. Underneath the hood, there are 1,000 microservices running and hundreds of thousands of data jobs.

- -

With such growth came layers of infrastructure and diverse set-up tools and platforms for the different workloads, resulting in an inconsistent and complex end-to-end developer experience, and ultimately less velocity to get to production. So in 2016, the company launched a roadmap toward a new compute platform, led by the vision of having the fastest path from an idea to production, without making engineers worry about the underlying infrastructure.

- -

The first phase involved moving to Docker. "Pinterest has been heavily running on virtual machines, on EC2 instances directly, for the longest time," says Micheal Benedict, Product Manager for the Cloud and the Data Infrastructure Group. "To solve the problem around packaging software and not make engineers own portions of the fleet and those kinds of challenges, we standardized the packaging mechanism and then moved that to the container on top of the VM. Not many drastic changes. We didn't want to boil the ocean at that point."

- -{{< case-studies/quote - image="/images/case-studies/pinterest/banner3.jpg" - author="MICHEAL BENEDICT, PRODUCT MANAGER FOR THE CLOUD AND THE DATA INFRASTRUCTURE GROUP AT PINTEREST" ->}} -"Though Kubernetes lacked certain things we wanted, we realized that by the time we get to productionizing many of those things, we'll be able to leverage what the community is doing." -{{< /case-studies/quote >}} - -

The first service that was migrated was the monolith API fleet that powers most of Pinterest. At the same time, Benedict's infrastructure governance team built chargeback and capacity planning systems to analyze how the company uses its virtual machines on AWS. "It became clear that running on VMs is just not sustainable with what we're doing," says Benedict. "A lot of resources were underutilized. There were efficiency efforts, which worked fine at a certain scale, but now you have to move to a more decentralized way of managing that. So orchestration was something we thought could help solve that piece."

- -

That led to the second phase of the roadmap. In July 2017, after an eight-week evaluation period, the team chose Kubernetes over other orchestration platforms. "Kubernetes lacked certain things at the time—for example, we wanted Spark on Kubernetes," says Benedict. "But we realized that the dev cycles we would put in to even try building that is well worth the outcome, both for Pinterest as well as the community. We've been in those conversations in the Big Data SIG. We realized that by the time we get to productionizing many of those things, we'll be able to leverage what the community is doing."

- -

At the beginning of 2018, the team began onboarding its first use case into the Kubernetes system: Jenkins workloads. "Although we have builds happening during a certain period of the day, we always need to allocate peak capacity," says Benedict. "They don't have any auto-scaling capabilities, so that capacity stays constant. It is difficult to speed up builds because ramping up takes more time. So given those kind of concerns, we thought that would be a perfect use case for us to work on."

- -{{< case-studies/quote - image="/images/case-studies/pinterest/banner4.jpg" - author="MICHEAL BENEDICT, PRODUCT MANAGER FOR THE CLOUD AND THE DATA INFRASTRUCTURE GROUP AT PINTEREST" ->}} -"So far it's been good, especially the elasticity around how we can configure our Jenkins workloads on Kubernetes shared cluster. That is the win we were pushing for." -{{< /case-studies/quote >}} - -

They ramped up the cluster, and working with a team of four people, got the Jenkins Kubernetes cluster ready for production. "We still have our static Jenkins cluster," says Benedict, "but on Kubernetes, we are doing similar builds, testing the entire pipeline, getting the artifact ready and just doing the comparison to see, how much time did it take to build over here. Is the SLA okay, is the artifact generated correct, are there issues there?"

- -

"So far it's been good," he adds, "especially the elasticity around how we can configure our Jenkins workloads on Kubernetes shared cluster. That is the win we were pushing for."

- -

By the end of Q1 2018, the team successfully migrated Jenkins Master to run natively on Kubernetes and also collaborated on the Jenkins Kubernetes Plugin to manage the lifecycle of workers. "We're currently building the entire Pinterest JVM stack (one of the larger monorepos at Pinterest which was recently bazelized) on this new cluster," says Benedict. "At peak, we run thousands of pods on a few hundred nodes. Overall, by moving to Kubernetes the team was able to build on-demand scaling and new failover policies, in addition to simplifying the overall deployment and management of a complicated piece of infrastructure such as Jenkins. We not only saw reduced build times but also huge efficiency wins. For instance, the team reclaimed over 80 percent of capacity during non-peak hours. As a result, the Jenkins Kubernetes cluster now uses 30 percent less instance-hours per-day when compared to the previous static cluster."

- -{{< case-studies/quote author="MICHEAL BENEDICT, PRODUCT MANAGER FOR THE CLOUD AND THE DATA INFRASTRUCTURE GROUP AT PINTEREST">}} -"We are in the position to run things at scale, in a public cloud environment, and test things out in way that a lot of people might not be able to do." -{{< /case-studies/quote >}} - -

Benedict points to a "pretty robust roadmap" going forward. In addition to the Pinterest big data team's experiments with Spark on Kubernetes, the company collaborated with Amazon's EKS team on an ENI/CNI plug in.

- -

Once the Jenkins cluster is up and running out of dark mode, Benedict hopes to establish best practices, including having governance primitives established—including integration with the chargeback system—before moving on to migrating the next service. "We have a healthy pipeline of use-cases to be on-boarded. After Jenkins, we want to enable support for Tensorflow and Apache Spark. At some point, we aim to move the company's monolithic API service. If we move that and understand the complexity around that, it builds our confidence," says Benedict. "It sets us up for migration of all our other services."

- -

After years of being a cloud native pioneer, Pinterest is eager to share its ongoing journey. "We are in the position to run things at scale, in a public cloud environment, and test things out in way that a lot of people might not be able to do," says Benedict. "We're in a great position to contribute back some of those learnings."

diff --git a/content/ko/case-studies/pinterest/pinterest_feature.png b/content/ko/case-studies/pinterest/pinterest_feature.png deleted file mode 100644 index ea5d6257894681f6afb0f969c3d0cb5b4bd371ac..0000000000000000000000000000000000000000 GIT binary patch literal 0 HcmV?d00001 literal 9118 zcmch7RahL`wl(gW;BE;JyphJeAwX~l79dCijng!q#yxlv2=4AKL4yXD;2PWs?r_OD zd!PS**!Mo%hg;vOnyXgLF~`)m3R6>&$Hk(+LO?*kRd_9{@g!%SZXXP!rzes@^8S+` zhk=H|xf#>?pdwdCRv5)$I#25l&K*gL~a?9IS`b3>lE|H2CX zk62+DDA)w%0M&AEu>HFW)T|s}4$f8%j&w3If2z*U$xp|kY+?qn|8vaxr&RxrTow$4 zxPr~)pbmC)|H`#6W7=(aQKv|$Pw=}N|5Xj3f$M6@{{2%i7 z7nb+G#d1A4!}X_8{=dfg`{=0~|7`y=22U^l86>d%)AT@}#v~Cf8W{nB_Pc_tl$QJ4 zz9IU1BJG=Wpo^tjH6({%9K%l%2}2es6zdhq{!Hbue3y(EyPIysv}W`v5%EB3Un(lQ zudZ5*%tj0al9Zw^QB{nzfQwtTNVUku{c`dw`?t&&8QVLzLaV}qt%)8-m#s&a@a1GZ z0t^%uB!6Q((UEKk$T%TN8yW$_vp?(qRQw|u73#i%V8~_yU$oAr7g*9grY5HQ5kqo_ zlAh@+mdvKNhggr%BJ<%uulhqPy_TpcBCN3(|30)mzE6&GpwE3?B(WYBtW`jByW z2HV_x(=V7uVM3#Io`xGr4(>YMA$=fZyb94F1oKA%bQu8xLXChT5|pMqjZuMS?zL#m zI4;42NavX2E9&oem^JB9p#)Q04_UjJmI0HKqL!Ju<~A}50UD%Up;#_;;yvh^wM4S z4Qu^d#;!wNsy2>{#28Vwk4+t*R1I^7=pR6m!9-ND;E3n+tCi{xSEscjBdguI`|m3y zrs#m*%w$U>V@{1vBsp(sMurkXf})x{W+`V(^!Q6ZL3dvA!>l zZ@FZ;)>N0Z#2({pI(k#c$`P#syD~>nRtm`IREiMwvbE+}^Jh+PFvK+d7O}`F=u|vK zq37Plv-X8qabOO<-mlVxNh!P$uNdd%NZ@tZ@*g_bxKaXJV=kuOieBYt>Df`oOkXz2 z2mLhv9{m`3raaNF?PcGl~;xtWFh;9KUTgV=ljG#yer(y!myydaaN&hhyGqOp$RyNgfz*p%xSAq%nP0YDs}C@k&ZI{W-~Ph-^1?Ja2X<+h$Y{5 z?DkYM@jUPITv2VYZ$)!7RP4H9yRdItQynilkFYurRZ{!LS}D_d*ELl>GDgYTvV*tW z`;qmnp#N)B0&BGi|IeE&eMe56+AT*(Bzb1nGOE<>#~fB`gj0;1E>q8J={KD-N}LmEmbxqUz2tvgA4z;+VS%+O*yHwdcqj}wj_D?e6@XPH;ZW5Yt z{|=54>?d$ld9bOZ5$y<8tiNKT^P`L^5Z(fv$v>*(ajmI_s}esesj=60PRc_qY`T#} z`*FTAp6d1n%#zINwkDq@1e>&M<6-br^j_e$kB}6KP?=~fDVZ1uS7mD0uM@!kIaSZ` zh|lpm?<;R%Xmj<;f#d0S0_J0F_+WYfsA&GVt(7e04FdmtOmRl_VM@~J86!;b!)HxY zqE1W>WB9?%hgqIg9A0wBXM$tq4EaDQt9%t1M!hgRGb_28fnuO-%LPe)pMX#>j+XWd zc5Iko6@H&ctd!}2Yg#ynhl*L)y@#IZ{OaIZv&WxoWknX2jAs-&NukfO8~DB7{<+f# zqmQ~ZMb;$Ms5Xd7tmOTntz3<4WnG!~$l6F5t9mqup|qG`D&9 z6S`ab9B|n7Q9&gXRuIaH#5UD>;i|)5-XUrjBK3nbrnzT|Q%SsSp$zp)K)oGt3tO9W zs}Me1rOobH`n196fZ7&<+rrkWox&Nv#X%w>j7NrU10GHaNfY}$B_ejzT78+S$uK|CaBgv# z6lq866Kt9gg_VgXG-Ygfz_~;v@?nMt5}M^=I5FFGHHc1wmPI`aY>xJq%CFv;$W`n% zdRrmc83^@nd5Co5bx3AnVkW+?_;6%w`OSy@BegIKx-wewIm&aXvAmmlEV5%}!9rQv zd;7@U5;IxK47@2dqS*xgd+#!NX>&eJz(Y}cPiYMc;O%->>``zNq1Sr=fGAuErekow zKfUp4%a=4gTokn2CfWoU((m6mY*6|T;;9^PI_ZPweHm$XfkP4!`hjQ4+fB9e(+i{% zAWIZG!HMpxDeD=n6gf~#diRGaK( zSZ62f!xrk#Zwxk84@y{eU(+a7S>Rv>tS5F`$I0}zAf5Hp-Ao%rQwmpD%KFP~C&sxy zc$C#Iqx%QtNs~ii^BOyNd>5$zE_rhO?)hOMD)dKMjrglT#1YYV%@&N*l9?sy7+D!L z#VJY=HaFI&HnoUJa>yC|8sng?iL6|cUt2iOq_>eq?Y8y%13!*xXNQsq zR&H4^r-=w`Bn>DtzB(vAwtr=JZT%u36*;N{68|M?{-KIW8oM+Hqezv!{wJ-I8SB^BcwbPJGBlh;#MUZ&uOQR+#)#Ff=DGcDg`ZHbYN>(+h$VJd>I zZoG^U5~Z{9#AJpdtJ79``r{3|l4tpET>Ca}{k1H7#GvkIMUU{erLGZFM#qE86iaYp zmnJ?%;q2CMrBL9;MA=D6Q(OLM`8c{{74PeEf}mKTx{P@CSXT zLo*f<#}vrY=-DJcvE943*86&v`0hJ;>u+ER4+5ND@$-gX zy5%J9EaXrf*_$!ZADQq&F%HJt{nFc&wZ7+iRRB{gw5+=9>{p{_opWVD#@x_zauoDj z9K5BL13o)_uo>W#%7&1XWZxL$eY+Y-5=lf*u9g+cm@=U69vf%q zabLY!!(c>Khob_V$H60mM$NQ$7@kp$MV*6^Q?ug}iaGC$EH*5GZAyyU`2v!|qnl*2 z+$^T=!~L#2O)VO0NUz-sXutCvyxZ^=LrO3|$io*Ma~X2peC?NXU&CV@2`LFfKG`;V z#Y-_7K6lq4od$*x7ioNZHGn{B`;}nQgTBJ0ceD6qHXZBnoQ;`0v2=?JMOHFcS3Dn> zkZMLfN_tQLe_U&tbWD2YLZEIv)rkaPOMIf6ag{xn1GiCSOL28rZ1FlKw z53JSbY2`)U4Wr48j%$r~T_IMAR^rQfr)9q9r8R^xnPHhI=A7Bg5_IS@AP^QArZI(p zJ0(_3;#_0<@{oLd!wbl-dSi>ndq`#m?XXnF1t{6)KG<$r zS!yxy!)e!7p6m#kb{{ZSX`l`B*;gq1Bxf(G+-Mp1c~MPWW*Pp7i|q4t1=A6-A2W8p z%=F(nv@01R3dECta@bD!@C?-MXu%@Hi7huLTD5ZbyX|u&u5a8mhUVU*YaM)4|`w*y`N*&M9-7 zh2Fa$_x6AsfHg1KnVt%z&{b`l?kUs{ExFCO#cmg8bRjfQ5R0e=j=OOwrp@h8$79Bi z&6U+u61)xjjMbPwXPN!mmL#MdPec-GeIsR z+XWKcXi}L0b;wI8qI!890D>tGd-S%-mS(a=YEZ-Y4#ZjqZH@+ zNm;Kt2RlS@(fQ&b#y5m+FsG*3H=P_*9<@MnK@rh8JZG!O-Ei#BrfrGKs1K7#t8RqL zu^ZCCL>Xf}f4IAvNdM@bMr8G67e0wQRL%fGRk)cDA_HtCrx3_7H@cXL%e==HEt zT$05)yw4Ijo_JM|;cOxj(xi?DA_~?S8&m=m6r`^twb&C7GBn9r^bHal zrjoouL6*ZpmG%?kGhT5!M4qugB%8!A|2mA`nK31_hsbhtks87oAxLF|5C?#Qzzs-Y zsl2A)pq#-dnX_q_c#F&YQzb;xULB@59==^_{n+sk-V>#W*AqbluHsHvT`QR?fqbfm*DJQS zrklFkbmkjdEb(jaCx(}Ct%su6P?u3}SAv4#Ytm|*3dM$nPOT;bqFB*?VQxJ8_Nnz( zL!UEZCmsp~1*0KH@~22N6ksu{I_Hd5On0vBb_Bg9x?`|vkaNUX2K+2o+ zlZh@d7LrTD8@70GxODzypVuw8b)XxljQ`TuTPi=*$xLqVv+Gy(i`b%UVy{Bq^s1RR z1Ce*-8b-C3xk8~Ggvj4z%4q@gw7rhaUX%K}OOU=%e~e;Ut$U&9;m{H~zS70^@#ZKo z96(1+yMK|+QkQy(li?`F%ejGmhzf=wv&^^Y^3umkbnXIct(eS+oh}53jo$T8eRjFt z$<6G|S(FG|_O9Spym*Ux*R@Sy33u37(EBc#;EVeT6$oVGnGBxrud~=&7GhR|ey+|8 zq`(5QBFGLF(aaiVp}ejJ?upnZ5isIpo<+*!M{(M%3O8+x&e7d+6%m(rTJh9oy=U+T zr{2*H!#1~{@iWzymQC_vTQ24XxZm+I(N>78`-mOcbsl>}aW#dse0W}M^zG>-M#aYt zo|hO+8#U+J9rXML;3?{qfIzN{XP8?TAHKb!8yIDQwetvX8EK9;?Y5b8A`a~+RQ7QO zzajIb0)0~&fYLS>@6NVC4!qGkaIn~D=U(a4DQQo8uw6E`={J(%vgFS8M*K*>Pk~Fe zrH6$sVmLMg>XMltNERHupT9(}es$q5D!@h^QX0N`?PR$=ox6L#veq;h;M=_yUPgK!Fr6{!PoKC!TCKD?^{dYU1}xT{)b}G`-J!cH|!Kj+(7*2$pHSl!w=;TWnw|M`&Oq+ZA;jnt%}-9+dtDAr5#g~TrRYsA=aDgiMT zIp%;Us}?Q(<8URkyS*;~C)WhAjv8YfbmHQiqNhU}(ifFxusf(9Bet=jCX)4j!S|VJ z`5T}EW9J*mG~pnZf~?Ill~_|s^m9zAt)+^w`g}mM zD?j`Mcf-d!5N|h?MDtqFJ124YR|`o@T#q3{A^OB~vU~`{%k!=b7nf4&)3A7pl8gJ# zTN2J!p|`ol4yH}7ZQgPbF$K~LN%3fKtNk8UGW=QCH2>UN22t>*z(kFmnvaKOq0_Wv z!o65oL&4*6dJui&%&|7#3x)%hq;!@VfX3&#{RnUT#d*sf+7=Nvj{&wXH*&L$<>w*i zOG6ST6mf(pCZ~RfC8Y)pOj}DL(FGDJNQ4{^NS#S%&4JhZD_r`Xam3Q8+S3aOeB;U) z7>U}nm2y2XZgzG;U}2dB+9B^-(%a?)DzuRsCm#1u67pTt9Iv_>G=( z)wlQX!53jSTuP#vBT~HOnEchIJ(oX%U&A57?;X*|f>xKR#KH_28Nv~ym`cxwfn>lF zt?7kJ!FR}kzQjqu&!b44w9ibolh7pH?HNr1Ns(cLnp7KIxa36B>Y1ZNr6#i{hHjz= zloXLEL}i5Z^%fcuZ=4vPe@|I4KSCo!l*w0ah91GA-Sj8$b>(g{-KXzW&(hDCnqDm* zLX|(+1qF!A(sSmbYa#pA?(wN)ndFZ1W*+Ugh0l`5hG@44%FiER^)^$pPX&r6rhbL9 zmtGApP9{Sjw6<`{U z&}d~}-`Y_57PJ{1oBNSk%zr>Jh?6U>=yfRD7hEMK#6&c@O6bH!;&XZjFYXy-Ks2zx@{42nWj zfS&W-8QTX*$@iO1q(T#{S&_Bk)a;CpO`pOD{_Sj+m4(3x-hi?ww<=^N0taptCccsK zRP$iix^o6Un^{7*|aLsc-`ZC^+eh*CR5?lr0C9Z8o5|Mse)_Jj?%mG76II34?thz& zZgdt>`CY=lHLVXS*`j*bzKQTDuZFK#s1T8V4803JyzGwtMpx(fn$&^iXf_1MZj896 zPZz=fYOi`T(R^L{v2FGF8h^Sm#EQt>uj(X}xu!@HqA(l%K(5!&IwDpDUu7!Fq;}y~ zI-YBtTdhpU8tsjbqq1`G>c5~fw0gSymS)SYeq$f>l^%?F6<3x%`o-&`E*@^vKz^TJaE2MPAdSW*EJ0^j>r)J`b+9L0iOw-VbVd zv^!b4@`Du)@xC=o+X*~&WbI)B`fE)1bD|%mSFa|F&w@p&Db8%zTD`j~_d{Y+uj&eh$<7E@JbH0VW)fW91 z$&#}|S8Z!~^nxnYfCCF0sndmu7e~%AXnh)NpYM`#vV0};@0vy&ym?OR6t0h&e~{wg zy|!y&l5+MC2uQGMmz0dX;}&Awb5O;-Ur+rl{XREWc~XWnyWO?>KrC*6qV;00Yu-L1 z&WyjU1l`}AC$3VI@>O{7@R$A2mYeTcNAW#4gZMsdmzB4-qPK#JB8Kq`O;j0on$WRVq!Cd>3)t3{h7K=jscctDVViZ1;x!gi{i%3ECYFOKdm_FiVA&gO5jTSx| zOB`$Efn{}3IwHP(q8z)0`$E+KhZfckcZc#1t2gf6OG;eKsS*O?6`6jv%dN$$VWe}d zH%ZT_u2{)ly)))k8uwMGyY#x>=yEcP_Ex|(W}mR$9Wn_(3>tW&)U@u;xujB2S!mM1 zFlEgPwIRyVnp>wNPp-X)bhF~rBG&->j3hHxjoBmuWd-h;4z)^f_x( zaGK_PgX^m2o|np##>S8jxA2xsZDjp*1~izr-butlWk-DSL?-PyZgy|+yj!uvbALgX72Fe#!de3%^R=1=@gIi6c=XLTbVJ@5sf^8m(cy}k1G`4|s2`EJ{6UY~ zSmSaHMD5%S%hA8PaFHR(u`?a0jBkBB7>XA#yzjR!vXN5NejifxUO5y0keBCtug&kZ zoYE(s!lX&P%Dm*XidlhF5$X;n9Do(Ql#lPy0$Dg$s~SFFn*2IK=~b1L&FeUQ{ME27 zs(nJSp~d*3XfL&E$S=saf)9VlQbELD-Tlgn`0GG%?8-(6F^*wlVB`yj$(OSGNMyw> zeg2J@pO~u@oO<@~Ba76##siw-yGhSkf~Hm{N&s z4<+!@q)Fv7CB^!f@75llf)#74i)RHvh(DYl_#XELVPn1s>jEiaWtUGBrL!k&K} z)UB;#C8|e3x;083IDWe}bVv}WNRx^6x_7TB;?foMv+vKB5#&Pq$s@n#tA98k-aq{Q zbiab1KmfE_Dt-umHGT4BGp(Q0D7M~qQ>{6|csNMUvUyw+zeS%!n$D((W~Iur8;5K6 zwEgU>)ODbLV5kIcNu$uL1sb7rZmo77?M$mPbGwU^xfgl1^)t`h;;+*Zu3GAMdfBUT zct3GU9?L_v~{ zNU}_Ao07jN!;>&EFbFaKtpAnxUxEHj=8f%3P?bZ9;jM=fAZ1F38EW&Aj!u^8s`l3- bDj9+}Co@&exntg+KV}p_DzatL@BRM^K`Oyg diff --git a/content/ko/case-studies/pinterest/pinterest_logo.png b/content/ko/case-studies/pinterest/pinterest_logo.png deleted file mode 100644 index 0f744e7828cc2c0643135267fa2a7f1d5fb0813a..0000000000000000000000000000000000000000 GIT binary patch literal 0 HcmV?d00001 literal 9003 zcmd6NWmH^E)-@751cJLm@CF)pcXtThG#acMx5g4AL4ySdE(rv8cL@Y{cL^>*8+mzV z=9%~VG4uU??^<&oyd)K;kBQ@0(urbLn5fBitl@w*QpY7~(gP|im-!C*f zET0XjyPToBj*E@Emn8&@AZ6`h1*TDQva|(jgDtIn+?R5>^4b@bIK`u_5 zmVfzhdONv3qY)59CA?iNL5^T|8Y{4^y|WnINpm|LjlH!PodKU3mzt{#*v?+j7XsGt zRo4aiI)a3(=_JHyM7@Qd37o+0mNec@@0_8+-ePqB;1z!E|1}Muqxpx6yQ3K0KSdd; zY0}8JK)^J7octUhZeA`L0U=IqelA`;#jTz@U; zZ$~w?|G%q~)4#o;?%Lr0)%zcbp}Ibf!+bJ#U;1{a>M6g=HXMOLrHDu8YgN zzq6=m=i=@HwR3T$k&*c;Yl1Y)YL*~-=f5Ud{((|c6IODDx?4Jfz)G@Wbk97T_V(7o z0&@JaQgZwPf&zlF+}v`4(t>ilg3>@a9%&(dSw1P?-&k1}kcSi4+5K;<_5Wf8|10*d zf^c$ujw}m?*n5JlF6~!o|ZQ$Hy%o^~?wS&szS8wf^r~`2VXGz_S>@ zU)%ftZ1>+&&o%T{_us|)JotC{gPosi9`an=2gu`D2nf_)lw_rJy%!EmF)Z-)?y>}I znUp?L{mCaTs2M{-NRD zx6vh~6%n>z$2l)@+_ub%mqrpj@`S7oxOTl%hueC80Sqlu+g?iIg%@x%U^?}xQLFm6dRrt_awh12xS(Ew3DL@f2cDU*SMW5{K z&F|+&#NFPPw|V0UcfqpD&U1?H_4szAZjm!#s)|ZPjW~$G^Gz?FX1GJtM1p4&d+X{bz57v zO=~yf+n9t^a_iSrgEaW?zSr2yxuq4Nz|@8#KgC^4f}4+1M>L6KHu|E)p*Q=ZfgpDG zl*b0QF2Ufc-(L&ejEss>QYg%MeMFT-HD|}#Hdbpo$dN|YXd#2e485k%-xa-2qKP2` z=q#LG6_prHd&s)AGMr%}O;FzvVC=+CH?1}T&6jU#0xxSXF1Gv8*+1pynbG!S1rfcz z&i+bPQ7WQipFbK9Cs0iYavI%7kOIL-dUr%2^WcoY zq0dGmcHSHRti*bLDEfl_n{6iwV9s5^Y?7xzeX>qjtLSEMb87yLqC}J(U4_;1!O0(8 z>%E#uUXnk4$bqqK9*TkG8vB}|klC;4a-+p36fVOGS(7g5UAxyh0eFde!>gTyuawgv z9Py4%t=y1`#@<#>jw%f@6g$?$TzRu_B^8p#aS6N36CU-ds?AKJ%%s)9S8s+cm_8Fy zqWuY-hA13$k5c->m#3Of4!w432Cjh4t@E{>I`ICi9E*Z#gl|$Ym(x=X%7cgW3KXA1 znfYgRqtzmY{7+xg7{`OV;mjoNf@u<_Oa{jw6#S{^H*XBdSNs7Nbu7}_R%bYDR}Vfh z2mKKEDXowlli{y7sTYy_6YY8(5m{?nl@&+LwNZJQuQ`JDixUPC1RQseO|otVP7B;8 z^n%qE?X;|a4S3O~VDpL1b*Nf16GYgJTzwF7QJjvE0*F!iL?Oa_-(4(tBu-BIe7w1; zS&;8{gi+tEr1u{Zv7^7fu=?fWHQE7WM4AW%4wbc+f^M%r8TIJBlZ3MlB40)qe&U@W z0>5;pW~Nq7@5n2=(Te1^tiqjK4@|*}3o3(}6dZ*Niwa*Is3rK1YvPv` zHy0{|486u00yuCZL=N1`G?4C=%$kNTA8@oi)Q%5}NT4YgUM|ZkkITZdAMnaIB^7R| ztT&yyKlH><5=yvH5Iaiv+pQiu+&V|w2H5w{P;#X6)0g8x2Q3_(CEmKlAAQoQ>kRGc zn!laLyk792u$@>YH%sV}-d}6RYUp{vJEN4MdPfT6>nz-P%E!Tzdz=4JixJir0EA|K zC0x;IcO5@|YDSLKSJs(3{BdaH)%!JoDJs+0pu9BT3JjYTZxu(RO%vG*N0pL?oZY#^ z@$C=N8chFSBeCyd{8FZa@^C!KT9;ZiiR0%aq2Atq`G`kTA^1S4n;Q7LC@GNKigrpZ zU)_c`@Zduhqn>Wg9_CTp!z1m--AT8~^igq@Y%AVYSN!ayP;*}n&z_1)+h|+n*L(X$ zxl@tH!cW0X9K>ctGgD(fB5e$qT-fGaFy!#3dUyTWIx8rF-*7u9I^7_)4qG z7%P7}LGAmlLkEWxZan(~(0+l~Xhn>`l)N1W>##A=<751$2|4uHC2ZkdOk5H04;3lP z8n3yLk+dAn(>&goc6^topf3(j>APMrpioab&l=Yp?(HWnw!dDE^;YI*j*)kdx0AtF zmr*v-CuHpagLfXbzsN*FT+ep(sx#v$;iOS06t*P%yk2~IQsgN~RuTt?QUHnYrHGxB zwrI6=`*q2oUP;JvzRNsR46q)$A-F06xBW7g^bjG8#X`sGfyPk%fx@|?O8j*6p z57k;6pT$gr8O~Oxhs0wI((EMw1a+_7$7(EN&v0}+3%|Rt${5A}vXEl~MOlv9W)>Ju z#q*C4gd#b}SERnf*alLs@Vbkdt`0=-f7qsp&cKz_P9-`rmb*+Ia+dZ?tX?EhmbCPC z2OxH!BfM_X5fIsF(-A-NEZ#wQ(^tua-{Yf?f}gf-H%=Gf)&_gBy_hF%RIYzdzP@2a zx-b~U?TuWiE9rc*H(7Wso}EE8V_nG7S@HET?SZjK?Vb5KErMNq zckN8yPV4pX0=(xU-n8y%y1v9l43+$?rlv-^w>zVD$47oXl58#W2%;(D0|}yJ^(L}g zF9vdwLD|WWHqAs#Y{n)fRG)z0oZBn``Q!t#)V-z``~x*CHgj>Dh5EXvl+$}OLTei7 zx6P>)q_g9@E!8wvzqm}C<(onZ8e^&`$J0XB?+T?kSaEF;Pv&b1&{DpyYty_~Um|2e zq@wk&=DQ$%;ea!Cr#k(nV$hIdEj1((wdO~Se?^2XsakMUyG%SokP44##BK}jTwV2H z@pooc<%TO`cq-jDk3lK->X$>ie>SV2zd}*7DLqbLCK9%s?!L@vy_IRJ8ehEmVFL_z)7D(orwG+_Yh|<(v*zfH;pJz_sPs|XK1(-%?y?fX^Q44W&zZn&;!E( zPc?gsfa<5aPnxOWz*maK5k#vWq8H24KQsY<@BHN6?AsFyf`Ru+ZX5w8z@xi-mf6cX zqE!)7=K9*aK%7xX&+9G8L)%aC$fV+h9>el2&V&}gFWi?Oj+L0SUc9uO`qJ%VACWlu z*yOBm?&gUTrsdVm)hY8z;S~fa7Pw zmwQd3uFFc6f(wLf(-E}VQxw%SJ7xxAP`T6ld|G&k+FCd%j}!2qv2$;d-IaVY`z|Yx z+Rk2jgIo@E|FN(2TEbKi79(z|htM4$?VyxoMv0U#<}k#rj6~tfvW-xvEA}J`K zN7EVc| zDK`t(FVA`L`m2e@l#s+D`$guePf4+T6i`^XfR~B0*wtciPvmsCanEF~E5|ueIymfD zLw7U5oywuV^<}vidsqZp{1+}zyna0)dS_53N3f;^jvDq21st(XA3>fEiL?;j$@R@s z!xdq$8Z?{3M)gX8G-&Iit`Q+8msQp$J1abvyjk39y=aY-x2FpiIX`;1hYF=w)wdXN#qr7`;*LR3oHT_UUm5u>+;lL5pbZZwMIST{A@J;6oV;sBSM_?i5 zxW?iip@77ItS6W99h_NK-q6h=bDI@DQ&lyoSz?G(NFx<7I$;(?(4jMbXyl|0 z=(%JVt&Q`!SX}$88bVEYs>SrtvO0b->01B9&xbSDr6=`tY%^4Kpq88B#JDx4$qSXI zlBR0~jnI4qbS;;Rkv8jOZEx1yhC*s?Y$0`7;k2{}0ET zT{Nz5-*n@0{xtR=Jk_1I3arMo*$j?W|2ZM{yZhF_O|SL<&%wQD&-LtLkR|qcjX}Bq z@Ahr1jaEUw-M*PrI4L4v0JCJh2<5@er7PS4vuVAIEy^Xbl+ha~V)7WeVj1QT^dk-s zAF(<5e*QiO#dO)zp}~TEzAu*ivi6x>kC5Oj;=ARReJM1?L_E43K(_Q%7wIHfi*d zmnDv)?wi_of{Jn}unf_KtKZy3v8fTE%5(Tn+&{)7ls2s(y`E9T=b7-+Rp{djb#R_7{*2*PJW-3n}&rZzYoesZ249nY*swZ$e^8+v0~Ez zM$CGTT=lNzC0JC&)UWuL(U(vO|L}bf){W^HakiH1%gX|#0d!XiOlG0+!H-vKO~Z?v zt%a7Ix#n|3ke^EN^p=0)ghDpPnx97C6ywe;%3XxSICuReroOkhPOjSHO4Q<<{Gu1p za*{J=o4oh(LG&L@)sm2-0#xH44#(T7IM$_csfqQTBy-;W8tdwqIr`cjo;IC(DvDj3 z_ltEHkb4_B67Sh53okr5AYw0u+kJO&;XC!SR~{cSWmmr=nHM66p>Q{lxRqWa1!)eP zf2wJ*cQXny*daY=NeS`nT#9lcT z;>lM&Dg)&*iNcO1)x{eXZ&g6?oVsM%L~ZES_bWQ_9LxcR9c?*OYF~YJ0tl| znqG&&>%_i$j!09&yJd?{f{<;zZUsrT0=;2v(U2Ik?E=BKB$8e;H*VpQYI#7ND(X~H z#`Oc7yShZAB@t5>s3i6KN1vUHL#jDSP-M>4P}Rto>)3#~^fYtC*+Vt0wn;H6Nyd&` zaD9=#hV-amgtR#BPy13hc3+J6w9tt;1?QPm2POx(ihb}p4h9Ra9tw?UWjs~a`^OQb zsP42mOBpMXA>860{G8^M@E9~p(C5s+Et% zomDOnhi+iZ^A8=Lk!>a3bN+l=X`sibGCt-^6rpmAoP=OeN*4VYZ1^?54=44)mRp4W zykb9Prx-0`9lic@SMiucHF;@l0S?S4ED?aYcJFjlx>#3~;Zf)@U~#b~^_ak8nruhG zb_`pezNyS>0xwWpJWXrWQDX44ssb=jZX1ElIkzC|%|noan-R;{Z>e*FNSDUqs{5ZQ z>OR|z1E6BFCwN<r}@<5KdEKIr3q}E~x7~NMdeckb2g4#jG5n)?%F?{j&mIeZtF@MrQ73doVqz ziN0~~%F+}b2JQ2LJC96Or*?&8PnF#XTW5t24|AqI*-YeP#P`<(Udt>Jn8*}%Q6=V$`miXxk7BF1)#!l=py_#ZB3|{~WxUD3wz9dYOm>bylyh`3QqK zZBtXdM_Z(KDD>rxzK4(LBmlm;AMQ1%R9uzOcbuhZcncm3T`6g@kQ%VYCm=Qa%C2o{ zWovi1x9bV&KNYjb zto!?oS5U2JK0;&PnN;G<)Jwm5{~-wcW#<-aA=$t8P0x3;tfe`IQzY=tYb86mya~K& zYB+tB)AMIw_x>#wT@c#7L-NM-g9qa#K&mo^QHW)12a<9kWjv;+xaW>je*w>Lg9}Bd{_(L1|Aqlr1YhV_E)%h0*r*Pxk z@5wI>8@XF0fXc0HD&5YTI^bSamg1VV$j?#D0s zyA?BbI4E*a2|!K^ctNSJ6CY>fqm-NPCjP)5B^ixaJo zh~PGgnTKD7K(TgPT$zVs-L-yj*z^dfNEy@pQA!2AT$9B*O0ia;oSqn=BOYC&V2?v` zFg~ncJi$^j*h)9z%`!{czyIi zcz0tHZzP=tE=JSh5Sc^T*%dLf;`w-1F~$!O%t|jG?4(yy|MqR8;#)Y9*4%rWTdjq6 z52H;S_?!lwdOR5z`$N1<4sL05G3#wILKk;6N?UE25J>_^{U^{xiL(qt&d6#0}LywmtvWp(!hcEY{W*on>qW)uSN zV3AKiA2HA+UHkO-t^fE~P`l{`Jg0IjtE&&j!P5M zcfXq77D}ilYdk&(TXfD&I4stfK$2?dQ8c&11{=xkDp$>yw`JcngSkVox>g4R?t>pk z#iXuPQtmm`Ff|+6Z1G+I`_nbYg0B=4(5}>uu7Bwmd8y;T*iLTyH_Dn`F!koX%^t(P zLpJV+mAF194d3*qnm4F*8OmqNscyXhzog+I{n*6fcYOHktqUf#`Pp7p1~p&+cAM05 z(Wj8c%Ns>0;%Oj$YXvtjhlXHF{FHvrhA!Vo%B3{ZB4DPJ%JsOHGIi4nsjcfI0g}Wu z{;KKuIC`>v^ddctoS4&Qm`E&Ko>sg-tu?+f^q7}`@~ z5aFkMJk4#%)y2ZF*`(}dN2A-XemdWGl1?XIIdEA{G{*B85wHuuYI85)mr@F)OG~*d zthUvVfX}5em^v;EtR^u$W9OIc1rBIV)~Z|CTwH+lh;36bfPUrd?YQ<++e2p)s=@EN z)b&s`B_$J2`SWe!xtPz(tSY))z+UX!? zy0Q>2mBsVv&Q~B*Y2lffryM`uBr=glFQD$z?R25n$}HKUZnk`xy5yUb`{QvD0b>YM zm@lu+?T`Vi@E-3^Iu_$^_q*<>pM$Z06x4N`n0;7glY4Ar44+DzAEi&)KHv5B_4{rE z4)D>I+HBbSek2e-&Rao1;J!6Kw?0fP$BZD#aJZmhSxu$g+B;`=L-EYRa73upZoa?$ zC$S^${U7$8osXdKmP_S6`s?wZ81iw{U(B8E{2NU+_v~lXlBKlYEzn;_8(z^?hJ?Lu zm0xA~tg&3d<9xoYT4-wT@Ct=BxZ`nl*R90}HZA-jCs+Pk-}tRI$HS2^qR^TPabGmm z*|6i024SrXobvdZGW@C1 zSc73oJgf%pQ#ms$SA6d^KZ`5Kl>+;p}sc{@3p-N^ - I would almost be so bold as to say that most of these applications that we are building now would not have been possible without the cloud native patterns and the flexibility that Kubernetes enables. - -new_case_study_styles: true -heading_background: /images/case-studies/slingtv/banner1.jpg -heading_title_logo: /images/slingtv_logo.png -subheading: > - Sling TV: Marrying Kubernetes and AI to Enable Proper Web Scale -case_study_details: - - Company: Sling TV - - Location: Englewood, Colorado - - Industry: Streaming television ---- - -

Challenge

- -

Launched by DISH Network in 2015, Sling TV experienced great customer growth from the beginning. After just a year, "we were going through some growing pains of some of the legacy systems and trying to find the right architecture to enable our future," says Brad Linder, Sling TV's Cloud Native & Big Data Evangelist. The company has particular challenges: "We take live TV and distribute it over the internet out to a user's device that we do not control," says Linder. "In a lot of ways, we are working in the Wild West: The internet is what it is going to be, and if a customer's service does not work for whatever reason, they do not care why. They just want things to work. Those are the variables of the equation that we have to try to solve. We really have to try to enable optionality and good customer experience at web scale."

- -

Solution

- -

Led by the belief that "the cloud native architectures and patterns really give us a lot of flexibility in meeting the needs of that sort of customer base," Linder partnered with Rancher Labs to build Sling TV's next-generation platform around Kubernetes. "We are going to need to enable a hybrid cloud strategy including multiple public clouds and an on-premise VMWare multi data center environment to meet the needs of the business at some point, so getting that sort of abstraction was a real goal," he says. "That is one of the biggest reasons why we picked Kubernetes." The team launched its first applications on Kubernetes in Sling TV's two internal data centers. The push to enable AWS as a data center option is underway and should be available by the end of 2018. The team has added Prometheus for monitoring and Jaeger for tracing, to work alongside the company's existing tool sets: Zenoss, New Relic and ELK.

- -

Impact

- -

"We are getting to the place where we can one-click deploy an entire data center – the compute, network, Kubernetes, logging, monitoring and all the apps," says Linder. "We have really enabled a platform thinking based approach to allowing applications to consume common tools. A new application can be onboarded in about an hour using common tooling and CI/CD processes. The gains on that side have been huge. Before, it took at least a few days to get things sorted for a new application to deploy. That does not consider the training of our operations staff to manage this new application. It is two or three orders of magnitude of savings in time and cost, and operationally it has given us the opportunity to let a core team of talented operations engineers manage common infrastructure and tooling to make our applications available at web scale."

- -{{< case-studies/quote author="Brad Linder, Cloud Native & Big Data Evangelist for Sling TV" >}} -"I would almost be so bold as to say that most of these applications that we are building now would not have been possible without the cloud native patterns and the flexibility that Kubernetes enables." -{{< /case-studies/quote >}} - -{{< case-studies/lead >}} -The beauty of streaming television, like the service offered by Sling TV, is that you can watch it from any device you want, wherever you want. -{{< /case-studies/lead >}} - - -

Of course, from the provider side of things, that creates a particular set of challenges "We take live TV and distribute it over the internet out to a user's device that we do not control," says Brad Linder, Sling TV's Cloud Native & Big Data Evangelist. "In a lot of ways, we are working in the Wild West: The internet is what it is going to be, and if a customer's service does not work for whatever reason, they do not care why. They just want things to work. Those are the variables of the equation that we have to try to solve. We really have to try to enable optionality and we have to do it at web scale."

- -

Indeed, Sling TV experienced great customer growth from the beginning of its launch by DISH Network in 2015. After just a year, "we were going through some growing pains of some of the legacy systems and trying to find the right architecture to enable our future," says Linder. Tasked with building a next-generation web scale platform for the "personalized customer experience," Linder has spent the past year bringing Kubernetes to Sling TV.

- -

Led by the belief that "the cloud native architectures and patterns really give us a lot of flexibility in meeting the needs of our customers," Linder partnered with Rancher Labs to build the platform around Kubernetes. "They have really helped us get our head around how to use Kubernetes," he says. "We needed the flexibility to enable our use case versus just a simple orchestrater. Enabling our future in a way that did not give us vendor lock-in was also a key part of our strategy. I think that is part of the Rancher value proposition."

- -{{< case-studies/quote - image="/images/case-studies/slingtv/banner3.jpg" - author="Brad Linder, Cloud Native & Big Data Evangelist for Sling TV" ->}} -"We needed the flexibility to enable our use case versus just a simple orchestrater. Enabling our future in a way that did not give us vendor lock-in was also a key part of our strategy. I think that is part of the Rancher value proposition." -{{< /case-studies/quote >}} - -

One big reason he chose Kubernetes was getting a level of abstraction that would enable the company to "enable a hybrid cloud strategy including multiple public clouds and an on-premise VMWare multi data center environment to meet the needs of the business," he says. Another factor was how much the Kubernetes ecosystem has matured over the past couple of years. "We have spent a lot of time and energy around making logging, monitoring and alerting production ready to give us insights into applications' well-being," says Linder. The team has added Prometheus for monitoring and Jaeger for tracing, to work alongside the company's existing tool sets: Zenoss, New Relic and ELK.

- -

With the emphasis on common tooling, "We are getting to the place where we can one-click deploy an entire data center – the compute, network, Kubernetes, logging, monitoring and all the apps," says Linder. "We have really enabled a platform thinking based approach to allowing applications to consume common tools and services. A new application can be onboarded in about an hour using common tooling and CI/CD processes. The gains on that side have been huge. Before, it took at least a few days to get things sorted for a new application to deploy. That does not consider the training of our operations staff to manage this new application. It is two or three orders of magnitude of savings in time and cost, and operationally it has given us the opportunity to let a core team of talented operations engineers manage common infrastructure and tooling to make our applications available at web scale."

- -{{< case-studies/quote - image="/images/case-studies/slingtv/banner4.jpg" - author="Brad Linder, Cloud Native & Big Data Evangelist for Sling TV" ->}} -"We have to be able to react to changes and hiccups in the matrix. It is the foundation for our ability to deliver a high-quality service for our customers." -{{< /case-studies/quote >}} - -

The team launched its first applications on Kubernetes in Sling TV's two internal data centers in the early part of Q1 2018 and began to enable AWS as a data center option. The company plans to expand into other public clouds in the future.

- -

The first application that went into production is a web socket-based back-end notification service. "It allows back-end changes to trigger messages to our clients in the field without the polling," says Linder. "We are talking about very high volumes of messages with this application. Without something like Kubernetes to be able to scale up and down, as well as just support that overall workload, that is pretty hard to do. I would almost be so bold as to say that most of these applications that we are building now would not have been possible without the cloud native patterns and the flexibility that Kubernetes enables."

- -

Linder oversees three teams working together on building the next-generation platform: a platform engineering team; an enterprise middleware services team; and a big data and analytics team. "We have really tried to bring everything together to be able to have a client application interact with a cloud native middleware layer. That middleware layer must run on a platform, consume platform services and then have logs and events monitored by an artificial agent to keep things running smoothly," says Linder.

- -{{< case-studies/quote author="BRAD LINDER, CLOUD NATIVE & BIG DATA EVANGELIST FOR SLING TV">}} -This undertaking is about "trying to marry Kubernetes with AI to enable web scale that just works". -{{< /case-studies/quote >}} - -

Ultimately, this undertaking is about "trying to marry Kubernetes with AI to enable web scale that just works," he adds. "We want the artificial agents and the big data platform using the actual logs and events coming out of the applications, Kubernetes, the infrastructure, backing services and changes to the environment to make decisions like, 'Hey we need more capacity for this service so please add more nodes.' From a platform perspective, if you are truly doing web scale stuff and you are not using AI and big data, in my opinion, you are going to implode under your own weight. It is not a question of if, it is when. If you are in a 'millions of users' sort of environment, that implosion is going to be catastrophic. We are on our way to this goal and have learned a lot along the way."

- -

For Sling TV, moving to cloud native has been exactly what they needed. "We have to be able to react to changes and hiccups in the matrix," says Linder. "It is the foundation for our ability to deliver a high-quality service for our customers. Building intelligent platforms, tools and clients in the field consuming those services has got to be part of all of this. In my eyes that is a big part of what cloud native is all about. It is taking these distributed, potentially unreliable entities and enabling a robust customer experience they expect."

diff --git a/content/ko/case-studies/slingtv/slingtv_featured_logo.png b/content/ko/case-studies/slingtv/slingtv_featured_logo.png deleted file mode 100644 index b52143ee8b6c650b4fe6c9d0b166b0f7163378ba..0000000000000000000000000000000000000000 GIT binary patch literal 0 HcmV?d00001 literal 8513 zcmbt)cQjnl+PCPviy9e3H)eE3i%t+VdN3HB8Acm5xeR^!EHUfq3$pYIUBeY44~!$wS(!xpf)JCUYHaH1{RltzLAHKrUt|s z?!piK%fs*Eg1ARxU`WaOAfVPxFb{wg%+A48n)RTul@;J%Bh6|kq6yYSD8TF;RDIoH zdcM!~t$m%WC2Uw_WdKq>kb40xFb^oe$Hmzd3GtC;{fAu0{rE2##0vNai-(gm>wg<% zq^S*1fV;y0BK)F2YXM;}Kum&PKol%2BESa_1Ph3Pz#<^9C=e_F0gFQf1pxoNSnsvD z+t@;M6_x*~>;6re)!xGc0Re%$y}kLph4|s_b|3)>2?-Eb5F{uFyypNSQLY|PAD}Ce z?OzfUVMuFt2ZVFYj5|>i~%ZZE0DS{;wL=+Si1pkFqb47YUU9DmN z!aDp9R`GwuLKNI#P!G7fJ{<1+uk2~t!#&_gdpH81pzx1u0k|}w)()D+&m+{u68S zAH4r3R`LIe1>GA1`kUVWBi;WZ-Ph3H;lGRb{^j504|Bb*dH4I`-ku(2#K3q|rlu&T z?=!O-K;Um?h&qti;74u@%FOVAj$7cz%3hHi!P@vz#|ykU?+h7!V5<4F+F_CcS*nG} zN6FJ_bA%ljHLa-^=2R;YFUUXNzU zO7^Hm!;N-Y^EOJ`0vf~ZhorKGLf-`Sz3phy$Ri2|NLAwN;^}Mzu~d=KbFUW5El)f5 z`QB#b2zRE|=@cewHoO1u4i1tvj&n53J0q(aZfbL+yxGXgO)6s@j($AV`RiNu4F21p zP(Pn9J^r<02s&QF-S2~1LnE}2JaZg*7eA;jhLiYE7&=r?`USZ}c?OUj?_HwO2jX?c z+Ab*TllpBf(#w6@Mwj@9rT~%>a;I-b-@`0_F zRvd{3uAcZVYrS4jO_%X~6O|a@I0*oDKW}-n==%;hJFI0A;+RHZP|xfd^zhu(+LIQ) zHghNtYQzBz$A-llDzXIj&H`3enbF5URow)pr=hW;@841?Pv>1E4_SW#uS(VwBb*=j zZB>aQeYE*1$$MGyv8NXB+;Wg_C+XTu9$gS+TOG9>Ac4Gq+@zeA{MTes?O3F_zA#fd zMdlJGp>0&a=QVr)cq}fMR9fZ`sMfQbx+KQ>>6Ml%y?7XC@VX{r@##)Wvv`nMxn?=y zGi$Yb$qBn_6tA)ffRR4b!7GNPtqjvqOJy&;Jas!r&pNR+U(t9dF1YNE8k=|exFIJ z!NRj4ubFjiU=+?J+lu|GPJzY2=jiK?5}S7doEfXuLNek;kx-%6F;7ygn=WjBl@SSW zd|qV}WL6||mmW)zowJtWC7_Neqjsj^V^S7sXDItVRwi#S8XZNPO5m{^5cA0dw@pLt zI(S8WI);p-$-z!%pN>nO+O+@o0M@}qTzw0hayG@MQoD}W##fdsL-Hz9>~}H0f1)JX zx3=l_PmG`PHMz58qV^CW=%(P$z4z7G`v)US4-b{SLVZ+JCf_7=2*ocw$(%FR$8^3u z3vpyaZ3X=h8@kpzd?Bk6nWgwuR>3w6Nf8o1%v8OpY!cK<9#1}n3jqaS2OLT~Op%Ck zRQh!T;VB$WD;w_q8R;)mA3u#6Z1+3|2c%`ESuW!<2)ube7qK2hxHcm}arOnjJ#Osi zU`fl0{_8g^?zSBL*`c^=I^YeHY2cM*SFdzC`VOO&3e*@ocoVjiYPrmn9X9-S2It@v zE{S?A7|)KhiiUBzoiT!z*Q2!$BDPXcS3Yh*CBe|vzU`=s?>2SF0M-se{i>n=+7iv* zK#%zYYgo||22ucd`S2K(-YYfA{uers5e0m+TK#bj_aSN?*E70+d;<#MgePHb4ctaY7@3%N@C@~ZRg z+{4=kCr=4IWhwzrK+#3Ktj8u7&3~N3^d9pLzf@B{$!+P-M(RgU9AE!N=aV1_{QRTa zYRckx)*|N?>|FM$%(nzgPM&*@Jb=6>T}v7k)^r_R3x}dy%$lnWV!5bLCw^mSQgL|M z$a^G&kd&MP+XC|7utho9c60T&i3?2MnCj%)?)wZRz}hbkuL))RRZzkR9}E4H3diTs zK>H1HouckSubpn6K2Xb)+Wf1;sbd*;Qaka>hNeO90ezA{gBL=&5rJ+^h%0o3%WCKdqw%2>b5bjDefAWAOX6PkP?mrzLO7ynN1#wKaeoQx#)1 zp=wI*=gf!Nk2{Jw$tAD9WM$HtS$yiV3phY5sF=D-CQY`GJ#o7;{qNEXckHz0km-1$r#?F5p6SA}KsuW=3}O!qqNEZd%VJ{0eOcAu(Oc z=r|ruhwdjC!iLko$DlKAj3)jV`z^BxT{F@~6kZkE^DX@__>;?V0!Njf>%y3-YeKaF0LwGTjE^!~b>JjXG%NxIqW=_4|v1 zqKw*@&&7TCECIeG8zXf>Lrgyxg+Uk7^}*VAPY|&*ZRUyiJXE5><5oFJBZlWT%mI=_ z3co)nJ6Wpd2+(YqE=P;XvxL9XkV=_@Kl*Xo z(I%{lVOBme7reWzhQPhO)W*$tUj`x)bF^mOQZU{?nqJkO2SeNPsyWzS0@yMtrkk%u zm!$j~B?QN%&xBjC3C-J?CuM^NZC(hIXk;JMtv{B@cwts0NlN$t%aSjicI&L{OeTOh zH?5b#|NE<86rLW=^F=JYs`GvIV2ZarTo2wFX?M#y|Ge32(de_Xro5`nKTDwX^FpTr znk!~JCN<^3P|?gEi)dTHHORi(W4AdGq9l-BRejH&nf)gh^B=ddvjolOZgoW~U7uR& zlm-EhW|P+QEprrfzV?aS6_T!$&i9sw^nC5ICcLa;n+|g7jSs}#SDl$`nwoclY#WRJ zIvjYV7&vBpJrRgsLQ$|^NiQ{RWaKg2%G_(~vLE+ukwv@RXDNl-63!;2vHE`vO)?BkXg=`=lW>a89CruQ({ z2~o-2i-r^^EkojnH<_)dCp8<-h>iYUEhUv~!~GPi%IrFBVfuca(eXX+PbXhfgbj~}Jj{72VV6ssBD}osS)3a@ zS*NZw)q_XX4i}4tbfrC0^~;^r6XxM^KQf#5O8VNd(__<_YHjl06qfV`3p+Z4=ii+r zMjd}I+HA^ZqR-_NIShs;RWoCu;*JjcuRq9ediy1H7+SaU) z?@~({JUNB0{M9nyZ{km>GgIQ90BZ|+Aa6z67I19(c|P-s52{NmiLV!ZmI~>A4F7qz z?ZcukVINM!t7`*sc2x}w3We(!5pdJB;CDPFmvI>W&=B3aCuq48;(BNXIZLRXPpnYC z%5%Y7&eOf$7lQ;*r^9ISe^bei0u1p4elG z92iJ=rh(Dyw|GlTXQexR5ppudq<3IOX-=1zOGxlT!5;@d_;y2h&!_G3h00M|DE+!^ zetpZ6(gX8=98EA0bm83C*+{jkr$f$A^Js`rm8~aY>}EB1{ReecHPAt(NmO({Y+ACI zRK1&sV>Mi)rPYEa^NZgp$=DNWgM9ji)A-`tV(ab6^ZX}a(xHTa&S*}O;YKaw|N#){MJb^oqe%Sqc*1pNZ%TuNR4IW_WQV)OB&_qFrX0MG6i^elBNk>?3 z7H03oKv?~kDtqmSL8n(8pD54+Gzm*aiC)9qSg+5HQL zWleQr^rsay9gFt+mpB_LUoW$mff_`rL?Oq=Q4Op-Wl=n_X-~9{SqykmLBjX-PJMPuUL{5DW9ck5^6n@!<657`k8hY^t#<`;Y4iZqwrv%OJF zeL2Xq`^G6n88fsI1mjMuduBjjOldGEI5e7_;3=}r_`2vIO6Z6$jIp(}w{=T#x%D}N zFKo@Lp*pVK+)I5TP|_MMV|=}^yjvM{oGKB_yAu%HTgbRP|XvrxV zO&UbFu}oe0O}_A|pz3ko7Y1%=>ym9%eRj(D9~b*H3G5W?2N~0!HbGMB))=nh;Nx-T zm9a8TnQQlhlb5z$x?KYOUzs1>`GyNDI3-j*^D`fJqEw&)K!(}!ievrHE6B_F`0jqr zOs$HW!XKuxKWfc{zsqwtelbAbuZUMl(irfcT! z{1Ol!6r~tw?{ZqK&ISgJi2{?0;>?AYe!VI_wO8M7F!XICu7JhTj?>xFYy%Lj9mw|x zysz7G?w-gM5}8V9;9*Wp#Vl>aM&xSKfV(K~UWPaWK`>p1;KH2CT$2d5kCCge4+#l6 zE+LvU{jJLPhoQ6|ZA3&q>Y+zXJYSk!{^AQT%{ucpJgSeAAN{SNZ>I7#qxTVJ(A^~G zv8?^LYsyzDzD+T}R8$`kwj1hI$Wd>uwKx^xeVE)OTNF=Y#Kz|c`s*g+2Om`%bLDQ1 ze`}jApjxUf8cToYlwk3TTKIi~EkXf~I%iAe$=x}@1^hD_@T$`9xZG$+!k z((~*@THfWX;N12X&LHp13`bRHC|?X~yfmTJ6XnAX>8LB{x$b)1j+Uvi6e57iOGRyd zK-9@uXr%~wwrWRy`W$gYoT;yyBx)U5DXIvzOWJ)l?d-HzZ*+7LM|1d~;x~^mQ({@x z*Sm}M3*?8%`3iBfyd%d8!ZyM`o{G4LV(wC^6@4+{VBQ2Rxxx|AI(Uz@Sk5|wuVhiIwwi?S9T>t-(=X zUz4bhGBjLjR>!9jC>zH#S%GLA#S#D9+O zT-2oyV97pPwnkm_r^U@HO%;|R%I?8=k2EmTfMiyX5b6qTV>m+kO})gE{_Y#ld!G08 zK=XvXLFOyr))Rx$A&-bNm3Dqf+A!+YC&$7a^J`>cjyt~{2W`J$QruA1+gbc+0KN{A zH5X_S;=&?A*EFZb!?yxzX7RZ;ge_Cb{g>~zefNt}1Mzmpg-0akZv%^iXC{vq=a9zD z2P>L?ltdh#u1h!``l=h_f@=!wd>i%9b!0Z?n>-QC4@KIYTZ`mqiprVBu2+9XcT6^q zVo9B579zyFM=PbVTZZIZ>l_>;0fR2J)bXbM%*JvP%tbMOThV#%fVM z3}J8@kkAPyc0vWu&$xkmTle@{h8=!g5ky%&*M7hs(o)UMtNY7EyfJBlv_&=klKtQX zwdJFUNhg_vDs+1?NsDYcjZYEqap~K+4%a{m+(IC{uG&ZzLDTugD=p@nI05C|8_X;= z6%fMo7PXO|ug0mFm3AlcQnE?-I5yx5`2po>!ZRSvEt-0(>XA|;&DB*tg5A(@hPCmA zQyN)LA8jr)-pa>{#xEl0qdJ{b?})5Kqr)UB$J%?vJUEI=*BA=LpQ=1`M#s{Ote`d; zBP#_{pS&nVWP|vKud|GU$8f@vujx_{4V~(8>d9ZT@0JY<I+!@6(50b5*PiI{1pf!Kstyhv69mtUQ7CFRyD+!9r-FFCMt) z$nz8Eo|dz&b|i z2?QGPqM!Qp(`>wkDv839vXhrac@v*)-i^MRLKu#s&6F^X!}Bypn#6`A5YRyi%U1tg=HGs9ke^JEAXip!g($~f(EttwWPt|hfBF;C(zel*0v)k8G6 zL$K53`jHenXe9fc->&PVUwByur=I85w({N^-$jkgcbx4lm6A7y2_<&BMmubhWv(g0 zd`vMh1HL4LwhhIb-+>DFxOW7--zGTQhEjZQ;qvI`54fga;+%wW9T#KoniO{Y^A|WD zq$&f>yEDJ^n7F$q8~c z;Y%3vmSY9`l$d|HnZQ4e-G~bua;4ZvVvwUFP<%wuCLdgsCw)-80n|#yn{yt1@|VD7 zmp4)3O(u|@Bk{nvTswPvDy#NYlx{ehjHu)Ofcu3v;q;iwNfTonEe=69Z|)n*oTr{g zCLssT2Op%K(QCZ1#akoTO=2o>(dCfpg>T6HCUyynobo-bU&oiEhl`yaVP1!LLZUxx z%(Y8T=`Wo6tZHtCSf2rzJFG!buhMS<>bA-&l4-0KOb9Z$4psW`MkiXTGb`-NLvpLQ zPSjRUh~%jn6_z`z!0@V##y5{}2u~au__a=wWL-}!$R7uew@T^wobFwZMvcDkxV1(3 z$W5PU`rO^tRi2lYe4{t+%U_(!a_F6oO~dYN@I&P{VXkkX+xtGdJA5_Cow>oQvM{6L zvs=z=)C+2q(mi0Gv5TX%*cY`4ks00$f@hkPO0}FAphNX)qMh_XzBxgVj2!{zkPD7y zb%P{6uJR-Aymer?E z2MZE#+6c@EC64%FidJzAL$Xlw3+3Q%;;YhM6CXCZnIBq@)jaD7CV;rZpIq>2OoM>L zFU^V#{`Aw3%#7__QWDJ`Q}3!OqO(o9tv(K|OGck^_8e=Ok@wKNX9ssK2}&8>bZoSG zKf%MQg$81r4%|meM zK}KraJT*0LJn_mZlZLDaz+1GpMizF86NOJ(Ir-o86qW1P(pax1oPSIJ2Nuj;%Q@ZN ze@mX`xiLMy<~lqH3@kY`QomJkK@9~kFH^o(dN+ zoF_z7NRDTB>>D%Vez;p^3m0%GWt=0JI1lP}`LkSvv%Oov zJldcB;LEbV!!;E&nr|__#Hh@cDj7HN!Mi|`2z~sSkOw(cIJs2)XkglVPk?2;oNM;J zdbzp5;ht&LBG7qAnboh}^Uuc(zs!?l#tra!eYE;aEH+FHY8Rayu7rnP3N8O(VQxKm zYAdHos*1gNkviS5L71X@*OoD$epvQd?=hr*ribR^u-;C7cZZ~RmNY+hAbH9HVN0dl zRcHVA?G!>`|87VUqg>A|*JV*uc9DpLJG@3guv}?EI-zd}ubURLgnfy(4|PS| z)xN@woN*fvrdcR1C-o*~x`1q*;%YHJ!vwUkhtB8U?`B7xO5gCTT}p}XmcF~O&8mPD zmB#R_6;V&Min%CE=@`oSJ&3p17G1Xf@avL6EG$j;5_oJ=+4-=7-Pl%6Y=%3Fr~7t)XGDgXcg diff --git a/content/ko/case-studies/squarespace/index.html b/content/ko/case-studies/squarespace/index.html deleted file mode 100644 index 461e466d8c..0000000000 --- a/content/ko/case-studies/squarespace/index.html +++ /dev/null @@ -1,71 +0,0 @@ ---- -title: Squarespace Case Study -case_study_styles: true -cid: caseStudies - -new_case_study_styles: true -heading_background: /images/case-studies/squarespace/banner1.jpg -heading_title_logo: /images/squarespace_logo.png -subheading: > - Squarespace: Gaining Productivity and Resilience with Kubernetes -case_study_details: - - Company: Squarespace - - Location: New York, N.Y. - - Industry: Software as a Service, Website-Building Platform ---- - -

Challenge

- -

Moving from a monolith to microservices in 2014 "solved a problem on the development side, but it pushed that problem to the infrastructure team," says Kevin Lynch, Staff Engineer on the Site Reliability team at Squarespace. "The infrastructure deployment process on our 5,000 VM hosts was slowing everyone down."

- -

Solution

- -

The team experimented with container orchestration platforms, and found that Kubernetes "answered all the questions that we had," says Lynch. The company began running Kubernetes in its data centers in 2016.

- -

Impact

- -

Since Squarespace moved to Kubernetes, in conjunction with modernizing its networking stack, deployment time has been reduced by almost 85%. Before, their VM deployment would take half an hour; now, says Lynch, "someone can generate a templated application, deploy it within five minutes, and have actual instances containerized, running in our staging environment at that point." Because of that, "productivity time is the big cost saver," he adds. "When we started the Kubernetes project, we had probably a dozen microservices. Today there are twice that in the pipeline being actively worked on." Resilience has also been improved with Kubernetes: "If a node goes down, it's rescheduled immediately and there's no performance impact."

- -{{< case-studies/quote author="Kevin Lynch, Staff Engineer on the Site Reliability team at Squarespace" >}} - -
-"Once you prove that Kubernetes solves one problem, everyone immediately starts solving other problems without you even having to evangelize it." -{{< /case-studies/quote >}} - -{{< case-studies/lead >}} -Since it was started in a dorm room in 2003, Squarespace has made it simple for millions of people to create their own websites. -{{< /case-studies/lead >}} - -

Behind the scenes, though, the company's monolithic Java application was making things not so simple for its developers to keep improving the platform. So in 2014, the company decided to "go down the microservices path," says Kevin Lynch, staff engineer on Squarespace's Site Reliability team. "But we were always deploying our applications in vCenter VMware VMs [in our own data centers]. Microservices solved a problem on the development side, but it pushed that problem to the Infrastructure team. The infrastructure deployment process on our 5,000 VM hosts was slowing everyone down."

- -

After experimenting with another container orchestration platform and "breaking it in very painful ways," Lynch says, the team began experimenting with Kubernetes in mid-2016 and found that it "answered all the questions that we had." Deploying it in the data center rather than the public cloud was their biggest challenge, and at the time, not a lot of other companies were doing that. "We had to figure out how to deploy this in our infrastructure for ourselves, and we had to integrate it with our other applications," says Lynch.

- -

At the same time, Squarespace's Network Engineering team was modernizing its networking stack, switching from a traditional layer-two network to a layer-three spine-and-leaf network. "It mapped beautifully with what we wanted to do with Kubernetes," says Lynch. "It gives us the ability to have our servers communicate directly with the top-of-rack switches. We use Calico for CNI networking for Kubernetes, so we can announce all these individual Kubernetes pod IP addresses and have them integrate seamlessly with our other services that are still provisioned in the VMs."

- -{{< case-studies/quote image="/images/case-studies/squarespace/banner3.jpg" >}} -After experimenting with another container orchestration platform and "breaking it in very painful ways," Lynch says, the team began experimenting with Kubernetes in mid-2016 and found that it "answered all the questions that we had." -{{< /case-studies/quote >}} - -

Within a couple months, they had a stable cluster for their internal use, and began rolling out Kubernetes for production. They also added Zipkin and CNCF projects Prometheus and fluentd to their cloud native stack. "We switched to Kubernetes, a new world, and we revamped all our other tooling as well," says Lynch. "It allowed us to streamline our process, so we can now easily create an entire microservice project from templates, generate the code and deployment pipeline for that, generate the Docker file, and then immediately just ship a workable, deployable project to Kubernetes." Deployments across Dev/QA/Stage/Prod were also "simplified drastically," Lynch adds. "Now there is little configuration variation."

- -

And the whole process takes only five minutes, an almost 85% reduction in time compared to their VM deployment. "From end to end that probably took half an hour, and that's not accounting for the fact that an infrastructure engineer would be responsible for doing that, so there's some business delay in there as well."

- -

With faster deployments, "productivity time is the big cost saver," says Lynch. "We had a team that was implementing a new file storage service, and they just started integrating that with our storage back end without our involvement"—which wouldn't have been possible before Kubernetes. He adds: "When we started the Kubernetes project, we had probably a dozen microservices. Today there are twice that in the pipeline being actively worked on."

- -{{< case-studies/quote image="/images/case-studies/squarespace/banner4.jpg" >}} -"We switched to Kubernetes, a new world....It allowed us to streamline our process, so we can now easily create an entire microservice project from templates," Lynch says. And the whole process takes only five minutes, an almost 85% reduction in time compared to their VM deployment. -{{< /case-studies/quote >}} - -

There's also been a positive impact on the application's resilience. "When we're deploying VMs, we have to build tooling to ensure that a service is spread across racks appropriately and can withstand failure," he says. "Kubernetes just does it. If a node goes down, it's rescheduled immediately and there's no performance impact."

- -

Another big benefit is autoscaling. "It wasn't really possible with the way we've been using VMware," says Lynch, "but now we can just add the appropriate autoscaling features via Kubernetes directly, and boom, it's scaling up as demand increases. And it worked out of the box."

- -

For others starting out with Kubernetes, Lynch says his best advice is to "fail fast": "Once you've planned things out, just execute. Kubernetes has been really great for trying something out quickly and seeing if it works or not."

- -{{< case-studies/quote >}} -"When we're deploying VMs, we have to build tooling to ensure that a service is spread across racks appropriately and can withstand failure," he says. "Kubernetes just does it. If a node goes down, it's rescheduled immediately and there's no performance impact." -{{< /case-studies/quote >}} - -

Lynch and his team are planning to open source some of the tools they've developed to extend Kubernetes and use it as an API itself. The first tool injects dependent applications as containers in a pod. "When you ship an application, usually it comes along with a whole bunch of dependent applications that need to be shipped with that, for example, fluentd for logging," he explains. With this tool, the developer doesn't need to worry about the configurations.

- -

Going forward, all new services at Squarespace are going into Kubernetes, and the end goal is to convert everything it can. About a quarter of existing services have been migrated. "Our monolithic application is going to be the last one, just because it's so big and complex," says Lynch. "But now I'm seeing other services get moved over, like the file storage service. Someone just did it and it worked—painlessly. So I believe if we tackle it, it's probably going to be a lot easier than we fear. Maybe I should just take my own advice and fail fast!"

diff --git a/content/ko/case-studies/squarespace/squarespace_featured_logo.png b/content/ko/case-studies/squarespace/squarespace_featured_logo.png deleted file mode 100644 index 551b6da32119dd8701405202ab8c861d63966a97..0000000000000000000000000000000000000000 GIT binary patch literal 0 HcmV?d00001 literal 4539 zcmb_gc{tSV*C!MzYY|FgYzZ@F%rFLH8_S5vlBMu4n@pBrX6#9cB1=h;JzMEXNF=gk zP?oYpDWvQzWXm?T-`Dfh^S*!d`@QdbUGH4i%=g^qd$#-B=X1_|O^g-ZbT7XoKL-cL zUYr@$26(;$zTvz(fZrWX=~KW%f@$o?w57Q-0||5rhXI*Jq=0bVgwqro3V|H#*Gkdh z;NYTDPdG9iEiKR_nm2^7EdvSi_64*#ICS)ad(wx1Kt!S0Tkrz<-4N{Viw5XzHWUo{ zON8mE3;sJOM@uV^5sgj(At4&7Bp5;sq^SjgX{aHPFclD74WOwR|Orl6g~A|e zP;c+;ef=fPVA@drD~#VtGfo8iQlK^z2F;&N0?xx-_Lnk{yZ`KHTM&o_eVk4O4u#-_ zrIGx-DLzabRu>GsfsmQ5x^~=uh z^sz7l15Kocmb$T_F$`vm(n1-)4N;m%ga#Y|L%@xG>Ee7COo9)I^2;|B@cmO4^&fT7 zMsx~+Nu!^j(Y$_Dft3f1Nn?1>d_hJ=+oz6%AVEhg2_&k|Hd|@?sQ%tLmO`fnP{<~9 znm6dLT%)PKi9=lrg+;;*wbbDz$n8C#j5Rfk4Al*g>PA3p1}5O2y5!&F@29T8ztx2T z!9ceQ<-ZlqFBVXZ+vGnm02coMiQ)sGhYrBRYB_8efbK^e*5E`?|5QfcwFC#L=Zi<~ zV=K)gJY*e-b6>g*6>Yj>hRz$atti8%g zRqtG`X?|Lx96naP88rEREi7vxgOUBMZsr@NBoU>cppXb!?=b_5^MZ3h(a{mnN45zt zKR^I0fcSr5|KF^!(q-REVYWVZ=e@b_;818KgFy6B>jP(o>cZA?^YVJ~akjbGn3~{) zwmw5c;knI?wM^|m41LUOyzS;)eCCZvA=7*WNJEZxx6`SElO zZ}#`6Ga*AUB94!}-ajP0_&!wIzL!8C#7k*T^!Q5|S*9WG0Ri*y@0&dsX*bq(Gqz}E z)s9RaV=sPt5hpRaXu!`Zr}j8Mf0?S@bN|hscFMu?V{cbh=0<^pvrj=8n=7MNr?W2$ z9?sGZgvfthU6^z)pLmr%*^<~;ad47X$@Y$lO}^QOYJb0!$(6Xrc(XglKMbX;1LdRO*6z?@`y)*F}X+c`%)ubHE`S&qN~`6BZ?c zLp8xY$SJ_tfxQo2ml7I1`-)G7dU>e?ZqB_pT;h!JYZk$0psXJfoBJy~MP2IG=XbZi zNUBS2{3H%>)=%rZXM1|_>ziUuoNlD({nF%{ex491@13zHB?k2iNAU%e9%9k6ezQ}vms$hP6P~EbojCnsvhrS56+ot)4ZuSHNckXxtB3{OTIEc(dZufQY<#6E_zprD>O- z^fjvE@#}K9y0!1xNlRwd3e$vxn6pom#feOX7Qk7FbJhHKdkXb+evV^$56#RZ&j_}- zaOR52apala7V6oV+U0ZWbEc=ePnNoIj$T{sf9#!(pxT^{l0^A}XeTSq?AIE#1Se;p zPj_T=>TMkqndL9$3Qyi3@d-t7OO&LDcmbfxXQ9P!${!Oe=j2}8Qo&1W(dAOTru!@2 zn(a?RdW!jP4bbZM^6?G#FYb$b5#N<|ts_+56m=lyPH;wfxo00lj7!lh`Q#&aN}wzlJX=xXASLr8aORIK zb#l336u%bbyl8FcG9CbUP2nitj+nM%`ZfpnG=VdeLLjRsTp;K1;keykU)-L7Zsc_1bzR=*vspeQ; zjBHaZ`+S?HOI z`%&JbB`#yi7Yy~nH+8Dp)>k*zW*P-;yq*S3>!@=}L_L1#ykIkOZDImItEeKL`2qKB zH1)3D+LZc<_>H8jixY8hL&hb=Mi$FgQoiL|&IM5qY_7N}ww-gmtI?zoGCQEUw9h>UnY#`GCj&W4z0=vr z-2%=tiI3nee3t2_e;t4IqLS_<{%EN!t(eZhq2L$PJ*D2eW0RE~I!+b(n1sAO zQSD=LZX-K3Nq)DWpkl@f^Fync{hudVDf!mEhnk)#_VP#^f?R)iTfsA1C%7z@bM{S{ zs6&!~I=SVUcj1n70lpTXWcy9cAk6X7L9zs6iuLG%C~wNr=k7*NVL`eH;th8u=dwg* zP!!H?o@P2DeNAxFwA7#RCRjHp(yrAWavh86KudD#z3yy`;(Ou%58s=+v}nWzWGE+5-_RV zIG*!~D)bgwHwp>l4J$t-Ug1HuXBg&)k0Pm`%^XXdDm!)_PP;Tz>QM3rez}38S5+dD znqd>w}vDYg}1r`>ko*Aa!_dOOXrrNb> zBDBjGZ}1VooT$6Sn|lb4$y$3H*JHs+(NHN(A10p%om=+`)yQvZu!LXdH?@OF0&`g@4SxceI z-*d#KvR%VB{DC2Y?0%n|R)kX}4S+0xo<8lM6YF*t2WP)77+685YwkM2aBgqW$VeXHa8!Z8uEm-r|%__Zx59pSKUUb47ra zNOeA>tgTBQYl#(yriE;LYT$A|HsSf98abaJFVxj>$|M))#X=`9bLOaozP#~HbRDR7GyXDv^tvGfUGiyB!Byo{U(<^zk+XGiGGddM97TtiO$@ ztj$&8B!b^i2oAm{MXcJSWI@G3Bza`cxjce)!GOtoHP@?JhuH}d zrD}3=Se1Puk3E;Rve8Y?pBFnEJZsCi9|+6%nGKs8wd&W2wMwbu?E - Using Kubernetes to Build Tools to Improve the World's Wikis -case_study_details: - - Company: Wikimedia - - Location: San Francisco, CA ---- - -

The non-profit Wikimedia Foundation operates some of the largest collaboratively edited reference projects in the world, including Wikipedia. To help users maintain and use wikis, it runs Wikimedia Tool Labs, a hosting environment for community developers working on tools and bots to help editors and other volunteers do their work, including reducing vandalism. The community around Wikimedia Tool Labs began forming nearly 10 years ago.

- -{{< case-studies/quote author="Yuvi Panda, operations engineer at Wikimedia Foundation and Wikimedia Tool Labs">}} -Wikimedia -
-
-"Wikimedia Tool Labs is vital for making sure wikis all around the world work as well as they possibly can. Because it's grown organically for almost 10 years, it has become an extremely challenging environment and difficult to maintain. It's like a big ball of mud — you really can't see through it. With Kubernetes, we're simplifying the environment and making it easier for developers to build the tools that make wikis run better." -{{< /case-studies/quote >}} - -

Challenges

- -
    -
  • Simplify a complex, difficult-to-manage infrastructure
    • -
  • Allow developers to continue writing tools and bots using existing techniques
    • -
- -

Why Kubernetes

- -
    -
  • Wikimedia Tool Labs chose Kubernetes because it can mimic existing workflows, while reducing complexity
    • -
- -

Approach

- -
    -
  • Migrate old systems and a complex infrastructure to Kubernetes
    • -
- -

Results

- -
    -
  • 20 percent of web tools that account for more than 40 percent of web traffic now run on Kubernetes
    • -
  • A 25-node cluster that keeps up with each new Kubernetes release
    • -
  • Thousands of lines of old code have been deleted, thanks to Kubernetes
    • -
- -

Using Kubernetes to provide tools for maintaining wikis

- -

Wikimedia Tool Labs is run by a staff of four-and-a-half paid employees and two volunteers. The infrastructure didn't make it easy or intuitive for developers to build bots and other tools to make wikis work more easily. Yuvi says, "It's incredibly chaotic. We have lots of Perl and Bash duct tape on top of it. Everything is super fragile."

- -

To solve the problem, Wikimedia Tool Labs migrated parts of its infrastructure to Kubernetes, in preparation for eventually moving its entire system. Yuvi said Kubernetes greatly simplifies maintenance. The goal is to allow developers creating bots and other tools to use whatever development methods they want, but make it easier for the Wikimedia Tool Labs to maintain the required infrastructure for hosting and sharing them.

- -

"With Kubernetes, I've been able to remove a lot of our custom-made code, which makes everything easier to maintain. Our users' code also runs in a more stable way than previously," says Yuvi.

- -

Simplifying infrastructure and keeping wikis running better

- -

Wikimedia Tool Labs has seen great success with the initial Kubernetes deployment. Old code is being simplified and eliminated, contributing developers don't have to change the way they write their tools and bots, and those tools and bots run in a more stable fashion than they have in the past. The paid staff and volunteers are able to better keep up with fixing issues.

- -

In the future, with a more complete migration to Kubernetes, Wikimedia Tool Labs expects to make it even easier to host and maintain the bots and tools that help run wikis across the world. The tool labs already host approximately 1,300 tools and bots from 800 volunteers, with many more being submitted every day. Twenty percent of the tool labs' web tools that account for more than 60 percent of web traffic now run on Kubernetes. The tool labs has a 25-node cluster that keeps up with each new Kubernetes release. Many existing web tools are migrating to Kubernetes.

- -

"Our goal is to make sure that people all over the world can share knowledge as easily as possible. Kubernetes helps with that, by making it easier for wikis everywhere to have the tools they need to thrive," says Yuvi.

diff --git a/content/ko/case-studies/wikimedia/wikimedia_featured.png b/content/ko/case-studies/wikimedia/wikimedia_featured.png deleted file mode 100644 index 7b1f89ac98490d8c64ec722375bbcae0d8ec9880..0000000000000000000000000000000000000000 GIT binary patch literal 0 HcmV?d00001 literal 21322 zcmeI4c|26_+sDt?BD<_Dj4dS0*o`$yNF@6bVGK={k+JVvAwp6jOO!;GWZyz%%a(oL zLs>(S{W+>%-}!Ej=XqYwpU&%L=5pWH=eqCnzOU<^W6tXc)znZXKYIKq0088wDvDa* zdn5SskQfhqt#L5|UjY&a6+Ka# zvdPLEm2yLX0qjst=1?~~TYD_RO`7dTTm)$EHw&;qf4Df=NVA>W9}sGwt_fAZIHI5; z{GxnFxR5whOoAUSA|Wm)d=@GQgNq8lBn05Xd@wiyCV_y#p}!6`nWLa3wm+t3i4;KDqU=!iPFPR}|3{s}HH;Gmdkyn1L;mjm*MY$`QCI&x_CLoE?!UB^U5i*#GJ5ozPZ)EY6=I?f3la+)-}mf06C? z{FMDjg`X=V1)eKH!4YNdgmKitU~FZ6W|`)H5JD9c_OnV1%B^mWMBDFk<&hHjtLTpv z|05ns(cB3ov#;fYiSY@+bp&Au5n+Ug2>2@k2Ez&R)AKJe)G?N5EBC*}5JbR)5P~q= z7&xAPjR97WrMZ*&|0oKsFHRgwB*F^gXlL#ugSIobMhQ6BTT2Q2(TL;uJFgH57+Z`Z zI4zWnkd(kbHRGs$( z{(n2fKZN<)P%3CFn4#`CnG5FckKIfgb>q*i|LL$r|Jaor%pI|){hE_z`%lsR-o}3K zQu~@8wTm!E?(b(Z$o-N)SxO1~sr8R({Z{^xq5rQ7_%l9^{NKlLy@s;??+WT?s2^Vc z9t4ZAa&k3yM4ht+bK-wi?LXuGS@_ohqy+Xq0c_EKUW%Wse_o;g!lAPhVN;QOz?q`2LGl^Xu{QrzypO8-#m+M}Ih z1b>b8mu4K*{_gv8W6%aK6i_mvA|gUy{rp(=f4u`mKX&h)D;yMr!uEQZNTp9=3hqw+zgzIpK3zx=$_93nV5#c%< z;=-kIpnZtzKt#9>hq!QQ9B3cnIuH@A!yztQ8VA~kxDG^w>u`t*m&SqiA+7@v;W`}R z!liMbeTeHoM7R!zxNvD4XdmJ_5D~7!Aue1R2ik|Y4n&0OaEJ?+#)0-Bt^*O_IvnD{ zrE#Eri0eQ^xDJQ7aA_Q9AL2R?5w61_E?gQ1+K0FfM1<>bhzpm-f%YM;0}p(=f4u`mKX&h)D;yMr!uEQZNTp9=3hqw+zgzIpK3zx=$_J51(=pPRNqU^z^ z{#?Pw`&{wP=z)(0LXj$3>Hy%u2>^b90I;zKzJCS)XBYs?ngIafApkI7qOaDT2Y}=8 zs*2}y+y>?oJ-u|g3+sCatgQrv+{s89EukyK5P~#6mTx^~RGQCe3Gid|Jo37f1-g#t zSpTC*H5HBrl-aaNj0eP6D2PQU^C-jg}j;jm>AXAEMeW+zqOm!Cb3F< z&U4VF(Q|XOcw)JCkY+;Y7=Yq4Qt}YeLrY{wBv7E|FPL4hr8X8*?lA<62V|n=i zKGvECQD$l?zk0xqpFLp5uO3jculXU_*Zh?1YjA^k1f86qPq-T1@gdZxFpUY&63nk~ zXGl>P60YfyJzbc$T!~gPQoKCZ3a|k%AW~KiI9;V1FABRll)6f)w3m&&zNIcEd`jYR zZ!a9$K+Xu@;lYT`)+Wz=eM;}^6(z*Q(YN``$ah1oa6CMS26iv=n%8YPCVV-fuKULx zR+iruPn?JdHP}7zxrLYLI=LVu6mCXuKwM2uiyHEP;Driz*_B)mDM1EgW1e1$-R7D( z-CCq@(h`u1+LG@&{&C%&k$3Tt+1y-ukgo5U_SreRTR^G|bq!e#BY+Bzs|!z%#>!I? z)eaxCO}un_W~-Ry>qwp9_IEm8QIwA zTT>{9Q&T=@IAyfEawj1BX$QvDo6yd@)a1VI0z?Qx4+u98!1|7n|)TZn~Vs7>k&NQ%lNH zQ}(1$+DFmhhd9GhPj-AQD>|k7%It-w&CoGS6DzqNg!S|p;%Yex<)bVHqx~YLrC*t1 zWvNcs*RBNkRi{#P6kVt78{@$uuI_YjVj5D20iaOh9e34cw!i|FtuxZ@`?k0^ryOwI zC!-qh0|MnTZT3247b_aP+6ye|06*6ACg$>MH4&o=uG)@{0>-ZahKzGCcd4lquOyQ#TS?_EpkwN*}bFlbrj~RbeG9=u%vehS65K(FA{-zPzp>>{Y>4b`_u}dTbW~vS; zYp1$6jhD`mHG0O)Ww3D;@!~0^dINxOU*Dc?&$hff#~^8n=pK0r_tstx%)>Z$G4FNG z+c>|~oI@>YW&_Ek+YpkmGX0U(lM-jvbg-lDMrqU}4JX9-Rllp?FZwD_pyjB51twli z;;$OgyI(Gv7ncRI5p}l-*9W(e0eI*f244?#JOfa8fucx9xLZ#wd2^9vN&1CLA0};$ z1%|xzEhai~hl{>^yVSl)wsy&N3%hlv(r3kHX=khxnP-N+C4=7SX#2>1{|?v5po>0p z{y~?%wR?*ZRA@Z*GD6465)j*{o^<9kuH0Tl-FFPOI(;hSzN#4zqh84&;PTP1@QEr( zlKqQVDnOI;NV>h3%;x{G5#OWi{`M@O%c&9v}*_svsmI zmN3!g^0HWW2r<)q9YsdmqaWg@DK(MWcbA2Ec#e%CSl8_OWcP9VM|sP4iMy*Br?`>W zQSaG{#(@OH#I$~RRDe9IB&YYc-IBt@ntn8tx--P$q6|Ki7EWNyP<9-lvtipOOt!Y& z3bVQ*B5eNgp65vBwM-EqydFCk5M}P%lmVYxv<{SIxXTt#ZDjTMz)-*>A9KX&GiVG6_p`(%X;G}bzZJlznj$X z)ZO`aco6v-GbkK(I*^?&b7|OX1FwbyJR>>@@k;1Mn2dX#kjbx89z;?v?Jh zUE|*PXdXRzg88%a1K#!f^QQN9`*3@r6R1P|ieb%fIi1F}i=yINh@0oLjD`p5dWBwhOS)w#6f*g}T5R&*8KH}fRf zxp5E$LGXww;_`yi%PVO7{N<_>7#&$I@I`Z)y*s1g%+}AVU(38(xhr9YNkWlD0jWbB z=9)Dr3;Q6w;vI-1T%-bPlwtUx9_1B(b5s-Ok}j@G#522#zastzUH!$zh@lpq@%Y6I)J5Hb$oK%pc>p(DoCU-H}eOA3w} zW#2xkGH?ggh%Dw%ggsYf4E4#q&vzqEHJwAL>1!6t1Z*M*E*W;$k2CCb5a37jFo`ck z?>Jj_FDG)U#T3!ds%}Q%s?8TEolqO}w1|X^k z*HYOC<(@Y9tt?3NNsL(ZC^ajoS;nxkF%r`NP)0HizzilVgz*y)Tl}7%@;vWp4_AQ^ zKS{YWgje!y!pPu>YAVSY8jtc$r@HS~7hlpGLF6A1l!r8wV#w?Lp(`t}&^V zk2nVM7+pDEK%e5!YA}*Ccf*4R(wKx1T(zemI7*dLH#9zZj7|Q-oljO2udmoK*2Yee z7V$gXVM!p3iT?}{PKi<&i0rBDX20@4(x8()h}GyO{M}Ih^>coCOhJkzQE`irg4kOs z*;5UlgpZkN>LvHTt*d4Jz9lWV9qB za#Vg7BsRcLarD-jKZowCyh4ZhT1al^S>cy6wgi4-9qUygr{X?qROReF5A$3XgQQNc zx7zHGNtO=izItmkn*v0m3aHH7)~YMIvij6zfG>-&>E|N*SRc38R8M;CaVdc(96g5D z(DKSQvb)(oWhkdY&x70I9o^>e)-cO=t%|oh_%)evd0Wy@Af z4z^a>Q<|1l$3FC)#YMo%2MT8NrfAdK@y7-#V^ z4_g*dg!f!|xRr@Xa>xke*br)tUfT$jJuW@=y07CEsKx}i(rQ*!I0wB_C|Kr{`<{sC6evO&v-b7Wqs1S%XOaX9vPnLO!A(MN0sSF z4n=x+k}bwbHA3$CgB=U*)Fcv7@-I^el~|HCQBJ{?yLTMRj}L#Si)|*sT=`^FY&P+5 z+qOn;z2|C#XscnNiJov1XRWBl8@Wu<4k_I|=Rmd1nRi4ndF=vTeJ+`X6v|nq{_scH z{)AfEJ`&$vmEXQ4PpKKb9?Nns*?Z18mZ)SYKg{l7+?(jY%(KN58{Lmsyqu?5s`NJU z>b5!ODD~gJ+W1OtQ(Pz!XXal(8#&smP#wU$GSmx+tRTP5c0{)+$f4+AsTxaE`bzPL z_fCk$Dc-8k2+G zq@-22q($jr`VGm#?BTuz;Vn2q%ENoaYVj1Zg5f0Kr&vA5+%7Ozww+L8m_aj<_uxYW z@{Xs?i4^H~Z?UB&@FwMjoh^=O61j@ob#t?cy`G~c$5Mhbtffi4mD9ZDX0;647rYb; zK1pm|(Wm$R{>+5cZrujj{_5p5>B`~JM@wX3Ru|w=vmP(M-yVEz3`k#+4&wvFxr^1k z7rjeOAgXv{Ck&9H%6{DqMDr`s^!GUO9SXuM=i#C^riY^Z%n2%TWVh=Jj}gttN9sf( z!&bj+rq@4sFb|iG-}!_n7YxRp&+-x1V_0@GR(c}E8g|}D&aXo*b8c}?d05E)Va$A^ zYfX>DQ*FI3pR9;qI0h3DJTd+Zza?sA$1yohThsSUrF{x&D)E+ElU~IYpEZ~z>9a=1 zb*-8nF_@C_^%J&@GzBL-{@%L?3 zaa(-2iCOZ-V8|Jp=j)mz6nZws6Do2`RGHHt$qZEIDd)=VTNyEt#kv96{Ao-Dtj}|{ z=KEHCGXf~+fREE=PH))(P5LqNN~zY;ScaiZOyG(abFl)Hvu{&Ui3v%j6PQVj4_u*$ zo``}Cc*=uc!+g#C_2o`iOIORR#@y`eNebnqk;*5kK}Djf(H$MBTRX zmum(*owa6Bk9)ZS3sSFI`%9n{cFWX=#bkO7Hxx{@}vv^$|~q+0N#-8nZ8p3T-J}TffQJl`B{R~#8ktEgV zo1KwyqkK4;Lqu5ki@wX!2sw~%)iP0KWo7l?!Gi&)m8GTa=<7tR-qow4U0uTHj;F7^ zw^m#^RR;=fV(n;2Skuzdz}Qv0-*OiyUs}}OURhb`=+Lro-n_>Vs^u|gSIl>%#9{o+ z&8p4r-0xGwtRW8{JScKm7>JIJCM2cT(bLN@D$LBwyUX%0KK=r6V{^ETQU8YzUmCfR z9fWKnX>Gkv9GCVG24e2tU%o14+xHSX(;6we>(&x4vVA{M(iM@|y1mUWY4o|Qm9bNJ z!L@2@ILk1_H7eHz zY#8{Agk9$QjxdLKUVqs%<*_jpOQ$p*EcNw+0?Ef4ty0X9g9{Og&F8T(r)7AaXJlj) z7CP^4Z`c=mg0FMExtMF$8X6m^ZN1gboH;X4ik(T2!0PB$uDs{Xf3<&bsFrO^Kayr{ zdujfbzw~4D>Jtgy-R(Bny$bNi%9@{jkt?$|#i6?5QMo$ZqD_mc2_UpDmZUw(Ie((l zxw53BSb6q-cAvboG|n}HMPs`GVRbMHnCH1wy9i&@#JkYjsn_D`teL@Y zI2DSA9M!b7?>$2wJAH91w>e3=JTvnsa62sQ*z0dWS55e1bLJ?5k;@ack z(WwcB4(H*Bf-C&JiLN8)Hr2s_0aK69X@xeuBR*RT;5<|{(mP?{L9n5v^}&?zY_LjF zQ`rWW3JMCY#ar!_xh+r7@#;x!e7tu_YB}ib_~RIQditV*g1*rmU40We3a-UPjGZ0d z*|P=IWWjGI>+TGe++d&SgM~{jmaubi)x*NUaIsB=u1lu7J@ftfGtF_xdkhr8_2EpN zBO|TaCzKLd}oc+lK;8rp}Z)?)KkucYAm)&!1J|jDOXCh4C7WoMxU~6sR9tV$z2pw>S zmscSrs%?gsk5Ag^6V)YQ9ew@cq9Slt=;`X3A1uywT^gaIqpS4zHbHt9c!}VJi5`ue#OOnbLke}UfbUavt4n2`#1u&#$%?d^Bto**=dy;TU)0`NBi@w zMbzUR{bn@4V#(P#Q0~0#l@Jx zqL$aGiFky-1s>T8y4^iJdOmJ0E}~V>%^7QJYr#*4FXM07MrN(FbPF#f%Xo$M&aCR{ z>FMd}f}1cnJ|&CH$^BA!-_%rUI8fy)E08(GV1)U9K<{V0{;nX2AeBnaofz(9<(ZxxLKGTmF!ufI^|R*_VSXe0+RB{xk1v1$juY z>FJk}i_~6;%PB{k9KZkE(9lrS7c5zYcjp3$mKEK>wX)CYFG-oSjA7fmlEBJRh=TIb&kiRQR*#K_oq&aw39(W52suv6~f^7-s+g120P u6f1t*n)yH9fcf`ZGyig*C-Lf@B(U2}a^+FId;b1^yHr)uP|T4x3-~V@R@z(u diff --git a/content/ko/case-studies/wikimedia/wikimedia_logo.png b/content/ko/case-studies/wikimedia/wikimedia_logo.png deleted file mode 100644 index 3ad5b630342048494a4ce20f52ea46c310c431a5..0000000000000000000000000000000000000000 GIT binary patch literal 0 HcmV?d00001 literal 7875 zcmb`sg;!MH7dAeWj6)3|NTR26lB-$~#`gK>fO z7|~%30x`v@Daz{yr0-<}hEOTh^`epUp~<_}YEyQE@~k*XNa!>^xt*eHXQ326Bb2e8 zmeER4g}`AgLXtE-@+GyshfgMMF3uCSY%m4XSXusEEDmeLyIW5GEo=~4-+ys9@~yiS ze`9@n7Fo#tp6U2}{qcZ=uXYCZog52@FeVnx9*Rdw`WC{-kH~`j2v?5mAHYvQbR5qg9GTli8j?4>TV4G*?W6ZI&R27b;m$6U|N4R zwzxpKB1wgd0YO0^$VuS#uf2oDAqa|Y{+lg_VaV?UoR3Ku*p3H$${|k zp}3I>Fzn9bFXun%-IH!fePY8DH2qdT3d`up{Q z>k-^7sNdUn1~m6_R*FM;yJng9ns-)8c*t6CN^QRebIlDpbH&n`M| zzxy|%@QwdZ!X*#NNWS^ulkiV<&+phHQzD26`rDhcyJc_p=lBF%!?Aw1V^=Kkr{bK{ zNLaW~xvN4YSyJaJQ@h>qcV_i%7qv$}0_02ZML|$uK6V#sAt_~%RjP4%!wsjz#|Ffq z)_5XR5GA>TtIi&pN$aQB!&e-pGB<3Gv8Y&eWU$O}yZgR`PILa}bofIME3qk8p7ZB4 za)&>fau^Ut=V2c%CY&hGrww~1QJqL_cDlAme87kQ=x_JyNX*7Uygg%2IwUzX?B+OX zv8z-b+|^FOGVHr}?^494XxQ}%l5YS{3x}FM!sS*Rk}&mM!|LwRwcH)ae^-PcoadQKdn^2mmC{K>qxBG6`2)$k~}9@g zM<*f1{yyOW3)K4D#aNy6{lY(!?bDfGesyeusD+?DFG9DWR*%5b9|vepY|XZA$j(64 zlmfK|R{CswmnI?E zRw+98C=Oxt08Xr;IrxY9v8q;zP@ah({aVm^9O2m-ntc(n)N}qEm5G5iq|y?Tkw%lh zz=IN-f!Jj)x_?u9tlj)P(>UNGF`#!_7%8>?`25ZHXeSKX-@@bMR5us4;=WIvYqCZJ#$HMYTt+T27na^!g7Z7S7Ix z9hLVmb8nxYSW=9d2!U~izY{3JyBu1Xvw}Coewb&bsI4Yw`VObM2en_)L%MaeSn~#s zoSMMFmnZn=8uy~09B2b^k=94`@cxdTu1ZnKgx5TiUo02TM%pNqQLm;(`MYWX676v*YT3N__DzQA+r{8xWIa%5Rby z8t5o)a#{{2PQ5>h`ZsXuuzdfU`SDm=ExQv{pxYu?#tdgwi1ac8Mcjh5>s1rNt|2A~rK)<6Ts{qsiVpC1WlH6X8(GLo(ME6OA+WlU1ppe&x zv&az??vXbaRIkpkEa~>;(U;-czM>MLNX3n-I?v1PgxO4Qo;XXz8lVx zl}Of1=l&OWDFlNNQX=p{FpP*dm-d~w~% zv$f%xF}U2)a*{kD)rsBc^5DF!$*R@(g#4Dr%=4m#JQ-nIr!XOO;ygQrVrMa6n4|cH zH47&al;BgzRz<55lZbiszMMJnvbda{G2|}cX$_rc{wPAmioh2ab=)d}D8j{~3kts+ zlfDWF7@TbLWYZt-`-4DH#qCLlJ@-2j)hX*_p_FRJui{q*Ye0Y3Jo)lM2I)84dxNj| zb{R}O`%~+ii+%@K#N!8v*>Q9EFGJq1GDwSn`R&E-B08aM=tEJ8)ujmrMmYj7PL#NQ zK%z%8q{MQ+ftd}Ivfn`{NwJv}!Il0QdS_N#8q zE4^1NEJ*={lGoL!PZ0Mz;TL_44|sYbmw9oM@IZLWihYz#g-8GVm)ZE$(NLkJr|Ho*Qx85-XUo^f#6y82qPVMHe9VfigkdMy|#@Rr!zM5j`M;7Yh z?l?Lje|Ti>m)3n7dHq2c1C{G*i2Qgc_+i@5YNz^PVxDZ>% zClW2W1{ghC^FmiOCSf>jPReQ7TwO690W9_nmI z#;PhGkus`W;Ij9Z7ljB7AG#lAbW z=a+GbhVY+nbid9~_cZGZlJw=sC_$-Q5Tn2UU1TU%wv@sfZwMRX#TDi##G$L!% z+&D4NZuJrjM<6PwV17lc*mt)$vJ76aa#*zL z+R|xUQq$uP%FF;Q*MEB5zIU=346#-+{$#F(XJWy))P&o`?B&W_YX{ zPFp?%cd->wOW4QF21ERAsc6qK3FAYSeT@vZMDvR*9@|90408PR7n|sGr?wHQ#!$7n| z13%qZS;mx-o~y@xdQ+8a7R?{c!Z)oNX*3}I5NE?a!UI1o^BrBeQxo*av(O<(>9K(o zJB;wZ`Cl4oGU{N~>~OXO4xC2_O3z@AO~> zC431gnh(XHD=t>?#dGT7yBB2@xJ^tJG8{y7Gz0%-6pW@RiBVs{|! zJSLAoVFV-T{bAkG(k7M(?v}E#VN`ya8W`GHc#|0}h@h7!&edsIk#B#$_`&5#w$2KK z_SdWP{#O(4pn)@5R+Zn71>`s45UUA(cIiJUp{e&Uj z10vyXgOS69s7ess1x}UkwN8Y1b~04=gQOT_eO+~3v64-I1DdFUY!fGXyd*kiNGap0 z{#%XUk!Iitel*$1sy7fBhAJqrzxr(ujGAdQS%uq}$Q%Eblb}hEW>|}VCVl9W^R$fiY~5V>9j9;i>CcWoA4!Yq1)U8<({u7hp2o-}uL#px8A4zgFN ztC6 zVIkkMgF{)-h)rCCfY~%bV{uaj)|h;)M3YQrbM9yI=&~Vc)M>C)O$^P6BagfZtzLyT@f7?u*;I{aM zXnzb38B{2Oa5euR`>4b&2aAXmdY*?|U-$GjOt>@u*pTM5@zRT^_bw3644eA+(Q^1h z{D|L2o~`5`ySv@g{5~HHXVlry&+%7ROxGBUK6?c(*JKEIdEn<8^aDESt#Ra^F10o? z=Wi}LCE3Pz&@s88ed0X*xBm}s3ySft5$;sL#Yt)W^Qh6oD*Z-tjAlIHNw0gO-+o`h zOsH){iXLk?S2ZdBMXyDJ>z9a>Xkk!989TRIZ&X*1g{S8zM=yl8Ec7q?{T5q;2s`Ax zFJ?l?_xZ|<-wu~0T8H0|`Lz-iE=`Tyz%Y7U9?4=gXsWtu$EW*nOV54KNN}Al`RG28 zDpGUs?(jFdykMMZ7U%3?T3)9Rv_86C8Qp)Hxhb)*?k zOu#y`cF=%e5DF?o>Fj4h_=v2}?KLJe;OJ1+2 z>tR>I$eF8@?dKE&RuImO-~8ivvAq~GE`dQ}w)zi09SjHzf_|9ET509{_L%aMkU@17u$8)sygV|O>vRlP7xpw4;^h6(MlmDr$X`)yJkslkK{ z;jSJ|&Ulzw4G{c04g5+$%s&gQL8cNx{T4!^{pixcRs7;SLxmWPvac%*66`Qa=Sn{Q zDVt$K^@8T?7ti{$J!Y#0zu}x6g{Ioj<;Rmg+AlFtz|P9JxJOgXB$099i2_o-)IPZG zB#uMIW;fgFd9lNF!slndmvPe-bn2v@SD|N>ea)Xwcl#jOsy=IRq~9k)DPDc zMm<1CSHtRalzs09h9cNEmKZD!sw~fF7jr+Edg8us0{Vz)E-h_+2)@6&G+bPdO8tS> zeXsbxzsn`tPpiE~jE+>&>QN0VbuzcJ&;7o>WA4>ieu*%nlJqOP@d(?RC2U*#U>Y6= z0&PA`_r~cU5Fx;g9Sf>ezqYS8m`}_1bNi4^r{Sz}#|Z3l5vRx@V>9?m|D=j<91q+Q zP)eKR20YHMag7S7H&KDlE_3~V`)uj|s{a2Sx9!`vom?+7=PGSuX7`w0_?7qy^70Ps z`H?Y7>bFtBOU#?=u-NUhS$F^Dd1GT^Dk>_4SHw<#3dzGs=$|JgC!4y*$HhH&U)Vjq z+;6*ftDg;BA53+ftrK+`wfmhd4}w4-Rn^tAb*{7Jma^BbyK{}gVq*SN71mo@TM3n? zJCg&e>E56SIjfV+5j8cn|3PW={#M|W?wvd^jztzvl$vsJ#T4J1@6QL{T>LB0`eK0& zZJ0xVE8XTLds2E=X+^O!U)MLY7m6#ayIu_>3p|GV94xeIeR&R2*U+#G+Tu`mY7e_z zSy=%w$cBad{J@b`@P=8IsUwbx7mLM8`fUC#)M01s?*GIKT(h;#T#MT2;v*Z!T4^Fb z_x$kshK9^5ZIt4v`SKv%-rk1g7DE!~{l<-}lP&fG>&Y_nUsA`!`8A=}XMjabW#uNd zaYD+296U%QGW6<1Og;L={}=yJHY&HqQQiV)|8LTn!;2!dctcki9(_V@iyHu6Q+qpm11z`mqXApxxd~t@C8re2tSH0J3EN z;lqdSE&qC&58AFLOx;D@7ec11?16_eDKs38oBzj+5odR%RzoKVgTV*Ui@CMf{d#8e zCzuL1;P5*kj>}>jqnsd(S-o4vR~QysW>gvJy?=h~?dA1IDOKEaC3-k}rwPi&#kEm8 zX+bY!)15Bi9d;9kfg+LLUS@@D{?4(y-8SS;jvq~XsS742eju@t?%k6zP9BcMzJC3h zKbcorI&|c#l3RT+0PC5tPH>SoFkLMQdAvSkSZXrf7J5Bt5zI$JeAhzgGfhdtqp7XE zKV7}?^CQ>G49Qx6IS81~V(2*uvy2oJ9CGxN%QEaTzidZrAs}~hr}a0jh*mJL=$V1* zv`RV!+8HJrJs#8vl{pGA!zn`S0s;sBdYQ={Lb6 zhs#~@@$q|ECaB|}R>0XperLPJfATxYWG*@Ys%1!yul~~nfK*kD()lQpjAX+>x{@!Y z{&my4x8z{Af$j%QJN)WfozX;(1uQhVZ^;$F;X==?J243f#Ohm-1Vmv+URdmcM;D3z zyl?X20ebDd%GLS7!SN;PKf}q>_v|YC=1nJ`3@wbCe4I$QoK^ZK7v@LVHf~NQEfW(H z4@UEe!buq=pKx-PnKh0+D;tq-P1Jr;YFe+YrNt*8u#xU)XBW_FQ9e2QRVlvO@z>Su z9T3y~7w2$~;bxz$X90&xLbg4-)76eZgpo-db|lsa2nY}gdaiU=RaF7g;}&MzI3|!V zNg0t6I@>yxH8e~eoa6mBRgvR6_pCVhe4pe&6o0bD9rZ(2Fza`#!+3$V&GF*Nm$J!0 z@T4g?yzc;ge{Cg51BuQIIZ_KeoOa9tdVII?m?$wm>k&2M+AtfdIo|kHjBN@!MWs>~ zy+;*|DCs;39Eo8Ln+nTQ{<6Efs}MuVC>3zn5rOBIXof`A+7EnAPfrJO1rY3N3)KvQ zs=S3eO(D|gyAUKYD`>kYUn9BgYP)#gGhZ;kWA1!lg7UrQ9RVF1A{?Kci8tPvsi{ss zs`*UoG}Y8bd|QD5&pzAR+v}E3%pzb<77GdrhSNly&vvH)B0A4E`8qf_z#c&!MCIq_ zS5+p}{xUp1c+&6P-yq+^kx&W5>-i#@8D#4+Q{&`6=NZ%T?$^c_S#+DepNYu~;9Oy0 z;lF`Ayu5(jN=gLbzhv*FWMyTAgev_H7C!GW0D%J#pxX}Y{m{s?sZ;NxaGj=Zz5j4{@|IT(Ax9AvzynMn`5m^j z7VZ(>0P>)hxG!LGe}6x3Sp%GOy?};#ybO~UM-y|k7O9fP5u|Q6h3Av~k_q$ zZs+|M`OF&LLA_SGUlR~fKQ*ezmXCU1P^zx0`#~fQ5CkwjE}87jp@u-~pRYL%{?DvSdQ90GYKnCu7Y6 zQXf1PFiQ4p$~NPLKQ(+1umjL*K)gJg=OrX0fcy%0AwOxR>&A$MvnP6#U}J3!Na*I1 zr=6XhJ#Xj-qKb-&auBNE948YUo%b%51_qprHfm}=L(k{^?{Ck40RaFwaP;2f)TAo5 zpr%G2{BI%npcXaG!_ICCDB82E5+^#_!O@Q)e0r~)oSdqHs~v`hGhT3EtP-T}Z;k;< z0P1Go1IhPpIUas-(6CCXHFkBnYJ$U`~Ki;reV3@Ie32 z(0n`kUUbq7FT1KrS0`z3aB!!rd4JZu{gZJA0;Z2ZbkOj^V29VYr6#qKK_@WVr!QUv z1LXteT}z7;DBar_%YMKL1t%pXA=C+pX$0LDS^%TV+pbW6ivUVwCq2N8$ji&))+=C; z326Wd3gsrDZJkUYb*L_Oaj1ZGjE$#DO~o!Jc252FW_-I|kpR$k{uGThc&@_ofl&Eh znK3*p2sRO-&78(a)bh0izreY5@iX98s6LU{R*0 zrvYW(1TrpQX_N1cA5eHv^O=BPK>ap2m?Yt7KB=dt2T1W;t_m3lSX7z7qqVi4A5H6A zxG?|;e)aVs9@#H-M95?b2n$y$7f3gGR!xw;=i-#A>au^E3y - Cloud-Native Infrastructure Keeps Your Smart Home Connected -case_study_details: - - Company: Wink - - Location: New York, N.Y. - - Industry: Internet of Things Platform ---- - -

Challenge

- -

Building a low-latency, highly reliable infrastructure to serve communications between millions of connected smart-home devices and the company's consumer hubs and mobile app, with an emphasis on horizontal scalability, the ability to encrypt everything quickly and connections that could be easily brought back up if anything went wrong.

- -

Solution

- -

Across-the-board use of a Kubernetes-Docker-CoreOS Container Linux stack.

- -

Impact

- -

"Two of the biggest American retailers [Home Depot and Walmart] are carrying and promoting the brand and the hardware," Wink Head of Engineering Kit Klein says proudly – though he adds that "it really comes with a lot of pressure. It's not a retail situation where you have a lot of tech enthusiasts. These are everyday people who want something that works and have no tolerance for technical excuses." And that's further testament to how much faith Klein has in the infrastructure that the Wink team has built. With 80 percent of Wink's workload running on a unified stack of Kubernetes-Docker-CoreOS, the company has put itself in a position to continually innovate and improve its products and services. Committing to this technology, says Klein, "makes building on top of the infrastructure relatively easy."

- -{{< case-studies/quote author="KIT KLEIN, HEAD OF ENGINEERING, WINK" >}} -"It's not proprietary, it's totally open, it's really portable. You can run all the workloads across different cloud providers. You can easily run a hybrid AWS or even bring in your own data center. That's the benefit of having everything unified on one open source Kubernetes-Docker-CoreOS Container Linux stack. There are massive security benefits if you only have one Linux distro/machine image to validate. The benefits are enormous because you save money, and you save time." -{{< /case-studies/quote >}} - -{{< case-studies/lead >}} -How many people does it take to turn on a light bulb? -{{< /case-studies/lead >}} - -

Kit Klein whips out his phone to demonstrate. With a few swipes, the head of engineering at Wink pulls up the smart-home app created by the New York City-based company and taps the light button. "Honestly when you're holding the phone and you're hitting the light," he says, "by the time you feel the pressure of your finger on the screen, it's on. It takes as long as the signal to travel to your brain."

- -

Sure, it takes just one finger and less than 200 milliseconds to turn on the light – or lock a door or change a thermostat. But what allows Wink to help consumers manage their connected smart-home products with such speed and ease is a sophisticated, cloud native infrastructure that Klein and his team built and continue to develop using a unified stack of CoreOS, the open-source operating system designed for clustered deployments, and Kubernetes, an open-source platform for automating deployment, scaling, and operations of application containers across clusters of hosts, providing container-centric infrastructure. "When you have a big, complex network of interdependent microservices that need to be able to discover each other, and need to be horizontally scalable and tolerant to failure, that's what this is really optimized for," says Klein. "A lot of people end up relying on proprietary services [offered by some big cloud providers] to do some of this stuff, but what you get by adopting CoreOS/Kubernetes is portability, to not be locked in to anyone. You can really make your own fate."

- -

Indeed, Wink did. The company's mission statement is to make the connected home accessible – that is, user-friendly for non-technical owners, affordable and perhaps most importantly, reliable. "If you can't trust that when you hit the switch, you know a light is going to go on, or if you're remote and you're checking on your house and that information isn't accurate, then the convenience of the system is lost," says Klein. "So that's where the infrastructure comes in."

- -

Wink was incubated within Quirky, a company that developed crowd-sourced inventions. The Wink app was first introduced in 2013, and at the time, it controlled only a few consumer products such as the PivotPower Strip that Quirky produced in collaboration with GE. As smart-home products proliferated, Wink was launched in 2014 in Home Depot stores nationwide. Its first project: a hub that could integrate with smart products from about a dozen brands like Honeywell and Chamberlain. The biggest challenge would be to build the infrastructure to serve all those communications between the hub and the products, with a focus on maximizing reliability and minimizing latency.

- -

"When we originally started out, we were moving very fast trying to get the first product to market, the minimum viable product," says Klein. "Lots of times you go down a path and end up having to backtrack and try different things. But in this particular case, we did a lot of the work up front, which led to us making a really sound decision to deploy it on CoreOS Container Linux. And that was very early in the life of it."

- -{{< case-studies/quote image="/images/case-studies/wink/banner3.jpg">}} -"...what you get by adopting CoreOS/Kubernetes is portability, to not be locked in to anyone. You can really make your own fate." -{{< /case-studies/quote >}} - -

Concern number one: Wink's products need to connect to consumer devices in people's homes, behind a firewall. "You don't have an end point like a URL, and you don't even know what ports are open behind that firewall," Klein explains. "So you essentially need to have this thing wake up and talk to your system and then open real-time, bidirectional communication between the cloud and the device. And it's really, really important that it's persistent because you want to decrease as much as possible the overhead of sending a message – you never know when someone is going to turn on the lights."

- -

With the earliest version of the Wink Hub, when you decided to turn your lights on or off, the request would be sent to the cloud and then executed. Subsequent updates to Wink's software enabled local control, cutting latency down to about 10 milliseconds for many devices. But with the need for cloud-enabled integrations of an ever-growing ecosystem of smart home products, low-latency internet connectivity is still a critical consideration.

- -{{< case-studies/lead >}} -"You essentially need to have this thing wake up and talk to your system and then open real-time, bidirectional communication between the cloud and the device. And it's really, really important that it's persistent...you never know when someone is going to turn on the lights." -{{< /case-studies/lead >}} - -

In addition, Wink had other requirements: horizontal scalability, the ability to encrypt everything quickly, connections that could be easily brought back up if something went wrong. "Looking at this whole structure we started, we decided to make a secure socket-based service," says Klein. "We've always used, I would say, some sort of clustering technology to deploy our services and so the decision we came to was, this thing is going to be containerized, running on Docker."

- -

At the time – just over two years ago – Docker wasn't yet widely used, but as Klein points out, "it was certainly understood by the people who were on the frontier of technology. We started looking at potential technologies that existed. One of the limiting factors was that we needed to deploy multi-port non-http/https services. It wasn't really appropriate for some of the early cluster technology. We liked the project a lot and we ended up using it on other stuff for a while, but initially it was too targeted toward http workloads."

- -

Once Wink's backend engineering team decided on a Dockerized workload, they had to make decisions about the OS and the container orchestration platform. "Obviously you can't just start the containers and hope everything goes well," Klein says with a laugh. "You need to have a system that is helpful [in order] to manage where the workloads are being distributed out to. And when the container inevitably dies or something like that, to restart it, you have a load balancer. All sorts of housekeeping work is needed to have a robust infrastructure."

- -{{< case-studies/quote image="/images/case-studies/wink/banner4.jpg" >}} -"Obviously you can't just start the containers and hope everything goes well," Klein says with a laugh. "You need to have a system that is helpful [in order] to manage where the workloads are being distributed out to. And when the container inevitably dies or something like that, to restart it, you have a load balancer. All sorts of housekeeping work is needed to have a robust infrastructure." -{{< /case-studies/quote >}} - -

Wink considered building directly on a general purpose Linux distro like Ubuntu (which would have required installing tools to run a containerized workload) and cluster management systems like Mesos (which was targeted toward enterprises with larger teams/workloads), but ultimately set their sights on CoreOS Container Linux. "A container-optimized Linux distribution system was exactly what we needed," he says. "We didn't have to futz around with trying to take something like a Linux distro and install everything. It's got a built-in container orchestration system, which is Fleet, and an easy-to-use API. It's not as feature-rich as some of the heavier solutions, but we realized that, at that moment, it was exactly what we needed."

- -

Wink's hub (along with a revamped app) was introduced in July 2014 with a short-term deployment, and within the first month, they had moved the service to the Dockerized CoreOS deployment. Since then, they've moved almost every other piece of their infrastructure – from third-party cloud-to-cloud integrations to their customer service and payment portals – onto CoreOS Container Linux clusters.

- -

Using this setup did require some customization. "Fleet is really nice as a basic container orchestration system, but it doesn't take care of routing, sharing configurations, secrets, et cetera, among instances of a service," Klein says. "All of those layers of functionality can be implemented, of course, but if you don't want to spend a lot of time writing unit files manually – which of course nobody does – you need to create a tool to automate some of that, which we did."

- -

Wink quickly embraced the Kubernetes container cluster manager when it was launched in 2015 and integrated with CoreOS core technology, and as promised, it ended up providing the features Wink wanted and had planned to build. "If not for Kubernetes, we likely would have taken the logic and library we implemented for the automation tool that we created, and would have used it in a higher level abstraction and tool that could be used by non-DevOps engineers from the command line to create and manage clusters," Klein says. "But Kubernetes made that totally unnecessary – and is written and maintained by people with a lot more experience in cluster management than us, so all the better." Now, an estimated 80 percent of Wink's workload is run on Kubernetes on top of CoreOS Container Linux.

- -{{< case-studies/quote >}} -"Stay close to the development. Understand why decisions are being made. If you understand the intent behind the project, from the technological intent to a certain philosophical intent, then it helps you understand how to build your system in harmony with those systems as opposed to trying to work against it." -{{< /case-studies/quote >}} - -

Wink's reasons for going all in are clear: "It's not proprietary, it's totally open, it's really portable," Klein says. "You can run all the workloads across different cloud providers. You can easily run a hybrid AWS or even bring in your own data center. That's the benefit of having everything unified on one Kubernetes-Docker-CoreOS Container Linux stack. There are massive security benefits if you only have one Linux distro to try to validate. The benefits are enormous because you save money, you save time."

- -

Klein concedes that there are tradeoffs in every technology decision. "Cutting-edge technology is going to be scary for some people," he says. "In order to take advantage of this, you really have to keep up with the technology. You can't treat it like it's a black box. Stay close to the development. Understand why decisions are being made. If you understand the intent behind the project, from the technological intent to a certain philosophical intent, then it helps you understand how to build your system in harmony with those systems as opposed to trying to work against it."

- -

Wink, which was acquired by Flex in 2015, now controls 2.3 million connected devices in households all over the country. What's next for the company? A new version of the hub - Wink Hub 2 - hit shelves last November – and is being offered for the first time at Walmart stores in addition to Home Depot. "Two of the biggest American retailers are carrying and promoting the brand and the hardware," Klein says proudly – though he adds that "it really comes with a lot of pressure. It's not a retail situation where you have a lot of tech enthusiasts. These are everyday people who want something that works and have no tolerance for technical excuses." And that's further testament to how much faith Klein has in the infrastructure that the Wink team has have built.

- -

Wink's engineering team has grown exponentially since its early days, and behind the scenes, Klein is most excited about the machine learning Wink is using. "We built [a system of] containerized small sections of the data pipeline that feed each other and can have multiple outputs," he says. "It's like data pipelines as microservices." Again, Klein points to having a unified stack running on CoreOS Container Linux and Kubernetes as the primary driver for the innovations to come. "You're not reinventing the wheel every time," he says. "You can just get down to work."

diff --git a/content/ko/case-studies/wink/wink_featured.png b/content/ko/case-studies/wink/wink_featured.png deleted file mode 100644 index 3c01133bef701699285d864d8b1924ad6e6b8e10..0000000000000000000000000000000000000000 GIT binary patch literal 0 HcmV?d00001 literal 21292 zcmeI3c{J2*`2U9}v@=vnY3wm(3}YBegCToj?36JEV;lP#ktGt6Y!x9{vS&}0@KBN^ zTO>phDk<6HH(EVY&F}f+`^Wcu&&)Z@T=#Wf@9X}&?se{SCQM87#7-u5CIA4iQ$-oA zO?rPs`b%KgO8Tz}e41{&%}9MN7FYe6)} z3wzo-l4=70(#Jg=u{c}2E2lZ$ir^pvnyIPBbEsD`5g-kPB7?Tpv)*2Lhv zZE=zopyRSk(w<0?fIZ$7%js!v=iq|$lmUI!MUp;m6hlCqU#qy<%7ElH5^@@7XmKhK zo$;J-VR0}{L{x$kAt@{Zmz02s330-pBH|FJBt%3E3>85_C6Q1O&hIafEEDOIw6ldJ zQX8%GJsruC0a?4cIwBzu4-XGv4^d&FvlT={Qc@BEg+X92FsTOE#mm7J>j`#n;rb@> zLkEp_!8sEgT?s@7&JA6xInm8k1_atj^z-%ox$GS^G=8RXaQRM;#1Z0&b%clrLm~F| zkUvGbxN75nHS&*EPHNrvr~MB?itcy^*FQ>;`i!#B zpYmJae#SVuIoo{&ZGnT}?eO-bJT4>#B7X$nXian_x>ytclJckfE5tt~b|qN;1>lC~ zuVHvk!e7ERJm12;j;U{5Elq-=p@BT6|^kLA#6dgK+2;XhkATA*yA_0TRDT#{0;F6*Wk`i)a(vZJO{2Ec;)BHhRPC*okP*4&Vhf6Aoh$tfD z;bMx43UZ1HP&p`EP6_%$epBdgvMLTPu2=^g{(Gwlq*lYk5$0l&A~>)m4h98_;c%8< zNefE}Faisg#F@j*MMOkIe@Jf%|4kZ=cP6;wEtH&z_FpH9Bi7jkzcFxSKtD$A&)M~z z=+~i-B>dhcIT5%5S`ID_heOdDk18RiAfX^CE(%9OmE+4TPi~}{j>O_N z7A9F2EQ$ICkWFR3n*1$dBhj}Ce}HX?7ng+NEa6~DF$odU01zX6vXl@5i(6QVi(0}& zpytr8z&G?ah5tQ=hc(_omb6kMz)%Dj2E~Yol9p+tD3oL(p`>B@ttDS;{awU`XzA*K zb;iqCktUh!x2-|z$4bGeps=wOAUOFousDLl#_}QXL-*f-e@}q3#yVKxEo31-0)G^c zhe$&&YMS*ZBBT-yx82*#D`%`_`lXo{m4}_V3eoV?{{y<{sP7 z*i@5TgLDS7z`A1plgj40n{+I2NK2x#J=Rs0V2`!JLmVBfq#?fwH`V;Pmm?L3c0^}V zv+%N{&E-$Uo1!*$@r`|#v=CswE_ou-kl)J4>lmy)L(hEY`oi~=Buc4dW zo~#IBI-@p+h=e<7G)(o+w-zLS9TBp+)e-+Zmt< zq!WnC#wI{o48I4Fd;S<{_$yKbPHN=WbYCMkdA>(}n=G4-!e7_HZ>6MzHtCEH`FY6y zZHfH3{rpdU`1MHtld5D|)Ffm;D4SGVl=+Z(skq31P&TQ!DDxrnQgM+1p=?rdQRYMD zrQ#w3LfNF^qRfZPOT|S7gtAG+MVSwomx_xF2xXIsi!vWFFBKOV5XvSM7iB(VUMem! zAe2oiF3Nn!yi{CdKq#A3T$K5cd8xR_fKWE6xG3`>^HOn<0ikSCaZ%<&=B45y147xP z;-bul%uB^Z286On#YLG9nU{)-3QgKn{L*}L8A_GF%q=<{@*Ux70 z4y518dXRn;d#k8al=OpIPMor~1_0pA0|5930su?vr1wbxzzqrjOql@y$OHgjAMwhW zf0Rh?N-AhMjOW1HBrnGo`fkm&6KlQY)_pM+kFJ0p&&D+t1+wtDo5aw;cv1}+x_1Pz zDf^41-hujPU8tO( zGkbGp9x+@gQ@Q4Ae4#i=?=B4@-w=>jY?#dj`5JW`eBj~ko+ZQt#92T+0Ab1W1>ZG$4vH<$9^*R zWiwc8K*W6vfJJ70QAXB`Iu7G2(`OFs@N4I41|`DN&+>Qe37rf|@joJ{<(jZa6MF+C1=lp&A{nqjXMd>G;4KH~hdcK#_o-db@5ddxl&}#droY zeOEbSZ0Sgu_oJczTG^Rr+fb*^B71`F^ zy;E|Da{9s5Cun@z-BwmsOz^BWIbEmbeVCU!B*N?8%)d07$Xu|OZgPdcT!Xa5eEBb2>|R^Yjx{9kW@qy*E=;vV+@w zBjZJ!%;A{>{D6WTi#$=nMcY~;CbRcB4tZSRA`oONxV&OH`i_n-a;efu`^7D6 z3|{FUbcOf6n`a-}4QJEnepQ!etQ|J&URfFkmM2=!)AVEf)ke)L<~}|w;b3j6QhuLo z)l^e>Mj>ohqiX=C9l$Pyhc!h~Kr~0KR3MS{WVvNHIEhovb3W_f`l`IG%}QB)*vt_R z{obonq0_+Sktl_iF&getG?}4=-JG~ca0;|Yxh0~W2&M0L;YvG#!Uw$ADiEk;AB~tF zxO_o>V$It+;wVEz(cmlHvD}AD{zv`2P6eDggMIP%*39a_(>L=cuC)sC+150=rG+Lx zkLKMIR?^t$^0rp$%a_`B?!6}3mUa4L1-s}tCk3xI!Yh|TM!(P+$e4m8XKch6J|ViV zJ=u~Ul=6Wwjgd2LZ;el5_71ygCM) z4nE>-vTsO(vD%t<|Eow$Asszn5C8qn#p^HOr!-PzuNa5&-c8*?-{@9<9<=k!K)liC zPY&y`+h=9GO+OC=R%zSXm?JF$-s!N^Y3Yqt3F08}2THt=D-s;WdYyq9dMEX0`%%HW zz}9M~0G3`X;lUbS=pMJA7vj zCf-ZO@C)x*v}CFe!}wpfxm0$%;3?$h``Ki+Du>f(VEjSum>T(KmY`39&qE_#g)W35fe{t@2!Cyzk<{u*+q1!ODzaH^5noQ5nq+F}Tn-&k4sM6{^I4uhp>+(! zSAUUNL+YCD_2lC(#kx<>Ny8t=@uug6a`J-)OSyXJ)cpdG#%l%5#~16v?0gkWw&#i6 zVu8Nex(B^j&e|zQ+p@;3XwI0sY7kfonS9LHU;ziLdpByBF$ZUOM731=oLGIT9mq-J z$((hU4tZCbsPYIGGimIj#4!j5V$gIEwrSRMJ1UsrWhfm?D?@oUofa;9PT@(vZ@HZc zFZ14=87ZO;5Uskwth&e54(|j5ob_<4U1I@?L7X@^&o9z=rc&b+vmt1rn{S3&;;^@pgEsR4MqI%MoR^L{`?7G!?ofEO>{k|VT1l@u;_>BlgMW@uyVd%csm}}51Mhm< zqHI6BTI5ZWt1B!pvrYH&#OEHda$Ioh3`6f>8x(K0J2*ccAC33+MV|4U@#vY4darD%&(f#Bcasyv+yS4$ja^&hH6G#E=Y8s$xNN6cBy-k? z0uR7x=c2|cS)}{&%t?kBUb2v{!B3+Ztb#~wf66;m$jX6MCH?^4J`UH z0imOUgMHRBeb%GKV~MB~=l9s->qIDPDu2qeJ&Qwot%_*xwFg#gldTjqJ%CE>7IL&w zLOr+k_hCct$`$I=YYR&98OSQjDjEQtd!|G05tF6#Occb@%;9$Se(lW3mbS=}Zd;+f zHtkLD`p|9CeQ-bdN?J|n&MT(OUHsi+uT|+|7|QhywyCp)C`mfGN--y*j^;-n*}Ff~ z_En@OvrHu2INvauHhdaoY@{dg@ogKA>y;SnrHQIl-`cm%6Mc$q6-O`dNBFPhq60PF zv&1Xi1*lraD$_w19?(a>3v{S>bQ^%4yHUH|Lv!x_+D<=9qBkL0h5k#@d4Ff)k!asZ zTaK_7nd>qfBa82&IZruUbb1kW^;4aheXyT-q@sQ++UXL9!-{FJ?1|A%o@nq&0c4=k z*0=NF`_GEArqkh9F62Xo+=tKlEJV8k&aj=n0R*V+xf@U)ed&-SpJi%`{Zqp*`-S(a zc>+;33{0K}Xz9~^gZUs)Ye&G{dIABcf!*tlG4Nplf~YlW_Z|Sb4NU*0e`=d<8DDUxY=kc$POs1yw7! zw?~**jS45hAF8!zXD?(IJ-5C4Quq{} z+(6NcXVlf4s+AreyBhwFRn$Fq06aU=mab8(eqt@55|iYuBK{%Ied41z?(D)?!b2gs zXHY9kOO`iu={%Qr`M;YUE#OtZm(j6u9NKKKQb)UjI)O~^^k7E0z(Q}l=vRIKyb7b& zlkR-)JGpeLe@Vj$Ja@>PF{ND>AaHj2E(0`hYO8vcGDFt7UCfIsJ1}qPbOC8b`7V%o z&j%`_lB;7Z?^n%D)u&F^#Z;b29*|__R`NY`*r)pOsV8}lQnI@D*K#EJ%bqaT4I45| z@8S_X#Z6l`tM1<%uULLn17@+BytO#y<91YPLlp*vl&DfU7J9i5J$3Lxb^Eyj@CwQ+ zdaF#ME|`BI*E>emebw-hQbfk#nsq)NW`KclezXP#8fPn{KHh)pJ?Pbz!{xiTaa_=r ztIY09A56yOq+yt%j<^rCp7P3W-e-8g!F>34F(T(F|pn;xFs%?)a4Rq_A)<0k4P zcg_9S4x_UCr~DP4wlVR;=yq`j24y8U=EDVDY;0&z4aPfl=ImvAg6Wz+?`T|#{YOl@ zM(FL)Fl*YVvI@Rr$x+|uIw&#jiCbE`f>TWLRin6wZ`^v4*vIP&dL(xD#QL)UJbDv+ z6TI1HOT5a&3!FWK5!|geb)SzaH$gs#0kY0Ec|3zjjtY>bs;o6*?u|#F0PUFM z51H%biSxSL{`ay%_DQ{1o-aW*f+WK{fVQ_YwgQ=Sq92Z1oM>X)*2nK5wH?69>svgS zawZ~(PAl?-=JoUPJM9@!0>M$e2dfiat_tW%>K|LU6wXC^(48A1bHoX^FHc!LAY*5# z-lr5nrwmDjN2r$}s_#a0)bgIJm%^NbF;?ya+^?R7vy>?dhlMLayDna@?HZ5*V%~Ga zJ`uW7&VR4AbE<8;_T-CQ?yQ$*dEOHWku+}h zqdwNK310fqFlM)yQNzf`f_*_K8rG1yA#CmHxvKkYY$}HX9hYA0oDb+kX(X)ssWuCE zq`DjD^{27KMO4>hm_P)?8G=C(!)4)nI;M}chTCYys>_r#9y& z*gF}AkvB}!{kep$NF(&86#^QsG6&JpztQqcf7Ehg)my#cVu5g2P4AfcI}6yT%2~If zY7MT|iELlZ_M|s%KpVR}yqvo2VMicS&x8(BRwTpyfoXjr$CY;^J$)d-W#2!go$^aF znY+(MaXHxsd*!yKl=nQimuIvfB&WR0QqIS(-*7B#Z@yqxe=G9Bo23_iobPbY-JI6P z;YG?fS?cs(vJhLh8y;r$TrYctJoAz>oX%g#z|$w96Z!U-wfSgqXr_NU-)!}Gv>t$% zlVyPGoHDk()X=c@>TP*?L1Dhh@wmcJk7$=hgE{%;!l>bVNIy4&8bg+EmcZQ=gJb@B zbxD&doO?SvCX{EZWNTv-uS^hM8{S${FRRaq0o)U1?u(nQhO>bgp4-U|nH}~B3{wQ@ zI`A!-CG{WmH%cCwh-5Jg+ZNw6p9k)0V$N!97c)E%Y9eo(DthZG6Kh#r=6rg=sZ5@s zCq~iSLPmuW`8+G8tn4s(z#u^R9@9xvwiPa#)C+H?1?QURPl4gE2N5+~tmd>8w_8RG zeRs`b@|RnNA}+O)d$={;vmfc;J&@b<$aU(WVFPtGk#CU{60a#u!n%XVA5V|V7^ zZ}e63f~Y89(&P4pU1TgQ5|r8bJf{3T z0a~m``w$4dATV=7K!jo4mA`6G|C#8tA}s)+H-(bz=FY`v)#iMNy&Wxs)G(=7j%1g~ zw+%{1tU18g_Is_<=CGbcIJcc*x^`Ss_ZCxNuiu?VL%!02+e4n_+~{x=<7o8joQ^6~ z23)<~U6|1-u{Ul#d6t!#7mw&>k73%4VAwvBk68GOq$%lQ=LQ%u0;hxEcq5*Zlau3u z@y`N|I;W!2Qgkk89q9S+s87eXi)JX{GiWdBi*JOUamzWMwJHIBS$*>clj#&W1KCHs zxZS4a3zFJLIUec^K2WNuKFF+eU2bH1BcTacF}Hm?SAhqfdybaR^Naw-fL0;7KI&ve zq9KGC;HiI`mvoo#N{U6k4g5xMH`(i(C&&SubEQ9~FLMFq zIa(N#^ch5BHf)GHd~(9HJZLb_en5Zm_1%XeHy%V)44UA%*+HPhXd;ivP9LTyp-Xnf zjp-VQ*U}$Skvavi;1(}yeT)uyz3&ng)lcft#oJhh$)%o7!=!-Xn@Tm!R_i@n>{EwNTXAF;1taFWm@;E_rd(&g4@%ABTnm=%x-^@csrhLn(Hnq$yyC{_|VRaXWq_V@?mc~XKL?!$f}8L=$SasZq>a? zYJHm^NH@V~Z<*FwW|z;Bqv}AEP>Ehqs!QjXIs2XVcI#+ptGa*S^5)Co-)?s|Zj65` v{C>Il_bcB&Ts!|uv``be^j@X+qCkKUDIp;Vgd(C4RFEcJ1p(Ldp{^Bi z94EdJ)TG4kjzv3^IB;S09I)0%FDwpn7zp^o1na8? z{C7|erWODlBpL>gmr{`Qke0m)P*jqVmRGteBPRinfk-QWAxdCrIZ24LGDJxkA`STS z0TR8TJ-w8zboKxAMVzStZ(^}1WiS|r!%5*}rI2ViSXxO*2@H_|%g9I)EhI4^{#adi-&;VS{+9K}{Fx?V!oa~$6j)jca=xX% z3{6e{XHx{?Z)*(J3ih9P|65^< z1yNK`(vnw{mey0esxPahBd?^WqopV#t1qYb7grbQ5r}~KWB=lM{)emhkKFTyK%j`8 zbzx}lAeg5<8i@e>v2kVZe~v}&ANBs^dj4}P^8d&M6N3Stm-oMv`>!Qp51qrmo0mBG zyZvGQ#GXeJoBNyZ+f5P@R$}(FY=WmYZiFQ9j)p{>&aQh@slm-ROGjg*k|-Wr;{a+J zb1+gHzM-)-+Yn5{Kc5$jDBLzbGZz!JR|=TtW&bQFT2;%yNa?O=MVib)4H7jptS+A! zoz1APlN}l94Lf?(>qMv@QLB4hXTN>wbFDBtdv*0}_WNn;LinlE+a)ylT?bnWxeKI@ z4>SREEPdMK=ZFKyOhmQJcZnz|n))0`=mLnylb`e)(XpJP|2xI~eN;Y;e`HdD$mZE$ z>|~ln+eB_p)AL-*bS~O0kSd)~U%liPt~%FdKQq|%=SgZtn=yQRL;0^MUsu&cvYGm|HJV9&-Kfcq40r6{qzRO z2f|rfMpSYMl#Bd7YQN>ugj8xMrUpKe>+9SW=%j`h+!UrqLS9DZ@z&`v&=+Rp@>)XeP)n%XrsQ^E3itjTdYX zOds!GNs9x(hxswZeVxf4a{?}FX_xU1D+XXv-%OA)s${Z692p&#R;|pgMF*xA@hUxG zSClUwj23`Z>X$F?(iB}*@*=%RN+3P+f>>O2ixyEfzK z9eU$AMFs~Oa#X8zIH!g1Y&^pbxo-3tNuT*je<<;DaZjZVgHufC)hu?6%W~F5kWEsN z{9uXHO7>>)fG+QW2~P&h`<&(~j_07i zqOd^)r-b6VjIt!G*)7jlOv-+^>C{Js7A31``XajV`Oe4WHB`%81^j;eI_`Ox@o8js z{~~WkDeJEKVeA@+3~rlObs=$@&jTmtEkY%cMj?e-rG0Ejmsw=ga;T;YVKGmb>=Lzy z#csxOhSoOP?s(Ezk=#?my?<;%mqvBMVwBLZUc+N>@z%wxTQ3jphG<3^m03nxv#Qqe z(RB#$oaMI4`bX^eaW5ut)=}))6K-hDt~E*(c7I34_z^%r6^L^?_~<9>tMp@S?G^f( z53V`ami$V~@3P^oV|{iLgLHAWKMM)MY5XMj>JEn~AK(nDnS*6$IDb#MRi)lLYOOz6 zc9TMaFP-G2URC!|5vurV<~0AQWQhtMElK7xp8WO1=Y_3NLmBpav!>VCd>7Zt`r#C& zwJ+!oxY48iXp;ziw%=7k6+ui-=aEugRbG136diGK?}~A{2dPn!u<}oSM%AXHWefLW zRN?P>^Y`7x(%#^I?D5m#trwoLYeF;&&o-?nYwthrqIWu3O|7hLZ@+FB^3vA4LLW-< zJdH9cXy|3cS%+LTZ3RE)egLUc*iqTe%$A>>#JG{y_3Z=BI=BHM+nQ?w)gEV{sQ=xp zZ0XhSJXHyN)d)8BYP8l)u${ zoEc3JX!5!~Z<7f-ipL$j%@T@`*loCao(Kgh9Hk#+kgs3sKAR*YMPHN_b(O|38zHJyPAuZ~{!9hA7r#j5>lY38l_fB$Y+{wq(i=NtFCwWdM zk1kx){R!Y7+VhO+G=|Jmr3_;2e%zs_5tgQY3zo0+5}H5@I&Y76QJ8s6ABgxhR#YYw z3vWam%5_EVfD_353(_I8JAqv-ba4P zz9QQ};1mebi?<9wwR~*og>Rs&aUfi>n^E1Wxps| zReVeryR8O2w;HwTkF)8xMjCJRv6aPt)RYj6C!a3(RVipz9Co>k1R2_hZ&-yZ*+`jB!izRu`;0#AT=Nh}Z~Vaw zz^9AjMoTwq zd+`9Zc=D0(6ZhphBLn*`5cQYC&&~okp^WlQTQ8{z&4~{)L}0?;a3_D z(&D?{a6o>lAq|mz`IyJ(Mu~)agyV|25Vs&#I52 z?Mq5HvX9m%6Azv+$Lf8r97!L#152lh*QU&sHAX1P8N7{^P1)ZN|LU3ZIxCuPx!&*% zyShp;E0WO}GqQ%$uZ+_pRc+$5MpoG2Y|6^F;?=+#$pV)j!5j2&_BCShgO}|OHJWQR zN^dWu2WuD4J>%;3?#7NJICPub@pjA)>sN5=$#`wb-E6fn3d7&ZwGFGVbseiuNDqJ2 z*gLT=h|`-0W77)>hhz{Y3?s%XnC@Xvd4=RF09JxajW8i2Oc*m?$^VdIT6Ui5Y}T>BdRNYxOeck(J$o1`ieIN#pc_5(KI zTF|I>+eem|>ZKPZM_hd07U#TfSdxn83@0*bwD}yIDA5jl%Q!LBHjIf59_(|NQc~Jl z6j|@J9z2X|oWK6kieX2+C&;eWmWig&Ud)QmXDqg8)$N{2ajI*C*B2qz zWYcZO60ybE&%MWwdM|S4_*nB9F!8?(Qd{C99@}K}$+z~o; zquZWxNh1`)Sv0fvPFwR@>WpP^lY2*s%dw=pYl&r}@Wh#N%dY%blQwnf?rD5zDogkC z3koLAq@B-F>--&WS|n1vGR}%-V@ti2%>`O%Z1`R|_x2JyonJ568nS%ua=%sU*jmv{?1BpK1S({~&!2609K!&CE0RwP$9-zy8a>oa}p}uCZjrfZO{1TEj)Hu7Z zPh3|Nz=hL$?>fmv4is)?i56n|SZec=zY|W5Vp%f3xoI1tveIVX{kofp4|j2aU&^I&{6|$VV4#_-1p5ZtNivKIoXp*=6=z&ss8~uTftiv*R0{!qSkyJ_7nz(wrG9#{lMll91l&h-KZGZ zAD@{{`)!dj#S1SC4r1<{x*BzpRi_gL_EV~r*J_($KU&jYXee02A3K}1Y%gKdmeb-K zG9<{SZxoLiJhqP0b%{2&ba34;U~9pSN{fnHZSA*X>+mVXBO1>fm)ywT!_hA zZ?JLCUUy|?Pt1UyzJXPDR^D=dx1n#D{ORRf5tUzI_dNTk*l*aHF)o`Wq(wlHLJri9 zoo5A4AP#-=Dkj~TgKg99|vdyKFiKvZtZppYnHg@c*QiW&aVzV=xd(# zsx*0sN$nRxv+of{ei8m;w9TT@pRl)R@i=3$#H6XsC^1*jMf>aS&4;s&U0tW>v$5PZ zwH!W)99Nent%ow(nkhX;?_Ib))d+Q5@t5)*BLDEobu-Q|1jEI98$qza+0QLFr<4Ea zAM$&6H{{#UPLr;j4eVCWsf&5hb(OS=672M|Piw$|KqQLc$Br)#XZ1q2Bux-JUm3>) zUIkCiSQon&hOVx}ee8V&OuuM%acAqD?P^%p{%YxV_kHp6hNt32f=;v~ZuzRUwFF&I zxfFt)V7XwLjR?5&pjy4>PIO;KVU*}%Oi)X{{`*O7ns6S&Weu0UgiQDB8s@6CGs?hy z1%q4`a*R15P9jx(Q#uqC<*#x^_PBZ05)AZfPpUR~Aicf!-cq|$p_H_EfCnId({r;U0~_4Gx24F&wgb$)}-^g7nflJbJ3a_<0mnAOA1{-ktUP&y%{eVl2?-}Hs#YCJ2b{Osx3sqw7f_>>R4&GVfy;p z{5K})pH{DIJ^0;UXkt@1SILiRPb|(V>YK;aIA#}1KYNUJZO%Fxxe-&ifv0lZPCUkB zcx7o!ayZYQu|rC@TzRz{fRwa7kvT<=G)MtAsoHNb^TLFhV5=ID6!Y;(`5igHiuRO{ zb_p(#Ln!%~*@brWq2I>_frxcp%iWHTtpPW0(dfhsk?U2cLkR!>o0R|E<;#N?3BTFh ZNQ6ZA;4Y$D!1JqEBRw - With OpenTracing, my team was able to look at a trace and make optimization suggestions to another team without ever looking at their code. - -new_case_study_styles: true -heading_background: /images/case-studies/workiva/banner1.jpg -heading_title_logo: /images/workiva_logo.png -subheading: > - Using OpenTracing to Help Pinpoint the Bottlenecks -case_study_details: - - Company: Workiva - - Location: Ames, Iowa - - Industry: Enterprise Software ---- - -

Challenge

- -

Workiva offers a cloud-based platform for managing and reporting business data. This SaaS product, Wdesk, is used by more than 70 percent of the Fortune 500 companies. As the company made the shift from a monolith to a more distributed, microservice-based system, "We had a number of people working on this, all on different teams, so we needed to identify what the issues were and where the bottlenecks were," says Senior Software Architect MacLeod Broad. With back-end code running on Google App Engine, Google Compute Engine, as well as Amazon Web Services, Workiva needed a tracing system that was agnostic of platform. While preparing one of the company's first products utilizing AWS, which involved a "sync and link" feature that linked data from spreadsheets built in the new application with documents created in the old application on Workiva's existing system, Broad's team found an ideal use case for tracing: There were circular dependencies, and optimizations often turned out to be micro-optimizations that didn't impact overall speed.

- -

Solution

- -

Broad's team introduced the platform-agnostic distributed tracing system OpenTracing to help them pinpoint the bottlenecks.

- -

Impact

- -

Now used throughout the company, OpenTracing produced immediate results. Software Engineer Michael Davis reports: "Tracing has given us immediate, actionable insight into how to improve our service. Through a combination of seeing where each call spends its time, as well as which calls are most often used, we were able to reduce our average response time by 95 percent (from 600ms to 30ms) in a single fix."

- -{{< case-studies/quote author="MacLeod Broad, Senior Software Architect at Workiva" >}} -"With OpenTracing, my team was able to look at a trace and make optimization suggestions to another team without ever looking at their code." -{{< /case-studies/quote >}} - -{{< case-studies/lead >}} -Last fall, MacLeod Broad's platform team at Workiva was prepping one of the company's first products utilizing Amazon Web Services when they ran into a roadblock. -{{< /case-studies/lead >}} - -

Early on, Workiva's backend had run mostly on Google App Engine. But things changed along the way as Workiva's SaaS offering, Wdesk, a cloud-based platform for managing and reporting business data, grew its customer base to more than 70 percent of the Fortune 500 companies. "As customer needs grew and the product offering expanded, we started to leverage a wider offering of services such as Amazon Web Services as well as other Google Cloud Platform services, creating a multi-vendor environment."

- -

With this new product, there was a "sync and link" feature by which data "went through a whole host of services starting with the new spreadsheet system [Amazon Aurora] into what we called our linking system, and then pushed through http to our existing system, and then a number of calculations would go on, and the results would be transmitted back into the new system," says Broad. "We were trying to optimize that for speed. We thought we had made this great optimization and then it would turn out to be a micro optimization, which didn't really affect the overall speed of things."

- -

The challenges faced by Broad's team may sound familiar to other companies that have also made the shift from monoliths to more distributed, microservice-based systems. "We had a number of people working on this, all on different teams, so it was difficult to get our head around what the issues were and where the bottlenecks were," says Broad.

- -

"Each service team was going through different iterations of their architecture and it was very hard to follow what was actually going on in each teams' system," he adds. "We had circular dependencies where we'd have three or four different service teams unsure of where the issues really were, requiring a lot of back and forth communication. So we wasted a lot of time saying, 'What part of this is slow? Which part of this is sometimes slow depending on the use case? Which part is degrading over time? Which part of this process is asynchronous so it doesn't really matter if it's long-running or not? What are we doing that's redundant, and which part of this is buggy?'"

- -{{< case-studies/quote - image="/images/case-studies/workiva/banner3.jpg" - author="MACLEOD BROAD, SENIOR SOFTWARE ARCHITECT AT WORKIVA" ->}} -"A tracing system can at a glance explain an architecture, narrow down a performance bottleneck and zero in on it, and generally just help direct an investigation at a high level. Being able to do that at a glance is much faster than at a meeting or with three days of debugging, and it's a lot faster than never figuring out the problem and just moving on." -{{< /case-studies/quote >}} - -

Simply put, it was an ideal use case for tracing. "A tracing system can at a glance explain an architecture, narrow down a performance bottleneck and zero in on it, and generally just help direct an investigation at a high level," says Broad. "Being able to do that at a glance is much faster than at a meeting or with three days of debugging, and it's a lot faster than never figuring out the problem and just moving on."

- -

With Workiva's back-end code running on Google Compute Engine as well as App Engine and AWS, Broad knew that he needed a tracing system that was platform agnostic. "We were looking at different tracing solutions," he says, "and we decided that because it seemed to be a very evolving market, we didn't want to get stuck with one vendor. So OpenTracing seemed like the cleanest way to avoid vendor lock-in on what backend we actually had to use."

- -

Once they introduced OpenTracing into this first use case, Broad says, "The trace made it super obvious where the bottlenecks were." Even though everyone had assumed it was Workiva's existing code that was slowing things down, that wasn't exactly the case. "It looked like the existing code was slow only because it was reaching out to our next-generation services, and they were taking a very long time to service all those requests," says Broad. "On the waterfall graph you can see the exact same work being done on every request when it was calling back in. So every service request would look the exact same for every response being paged out. And then it was just a no-brainer of, 'Why is it doing all this work again?'"

- -

Using the insight OpenTracing gave them, "My team was able to look at a trace and make optimization suggestions to another team without ever looking at their code," says Broad. "The way we named our traces gave us insight whether it's doing a SQL call or it's making an RPC. And so it was really easy to say, 'OK, we know that it's going to page through all these requests. Do the work once and stuff it in cache.' And we were done basically. All those calls became sub-second calls immediately."

- -{{< case-studies/quote - image="/images/case-studies/workiva/banner4.jpg" - author="MACLEOD BROAD, SENIOR SOFTWARE ARCHITECT AT WORKIVA" ->}} -"We were looking at different tracing solutions and we decided that because it seemed to be a very evolving market, we didn't want to get stuck with one vendor. So OpenTracing seemed like the cleanest way to avoid vendor lock-in on what backend we actually had to use." -{{< /case-studies/quote >}} - -

After the success of the first use case, everyone involved in the trial went back and fully instrumented their products. Tracing was added to a few more use cases. "We wanted to get through the initial implementation pains early without bringing the whole department along for the ride," says Broad. "Now, a lot of teams add it when they're starting up a new service. We're really pushing adoption now more than we were before."

- -

Some teams were won over quickly. "Tracing has given us immediate, actionable insight into how to improve our [Workspaces] service," says Software Engineer Michael Davis. "Through a combination of seeing where each call spends its time, as well as which calls are most often used, we were able to reduce our average response time by 95 percent (from 600ms to 30ms) in a single fix."

- -

Most of Workiva's major products are now traced using OpenTracing, with data pushed into Google StackDriver. Even the products that aren't fully traced have some components and libraries that are.

- -

Broad points out that because some of the engineers were working on App Engine and already had experience with the platform's Appstats library for profiling performance, it didn't take much to get them used to using OpenTracing. But others were a little more reluctant. "The biggest hindrance to adoption I think has been the concern about how much latency is introducing tracing [and StackDriver] going to cost," he says. "People are also very concerned about adding middleware to whatever they're working on. Questions about passing the context around and how that's done were common. A lot of our Go developers were fine with it, because they were already doing that in one form or another. Our Java developers were not super keen on doing that because they'd used other systems that didn't require that."But the benefits clearly outweighed the concerns, and today, Workiva's official policy is to use tracing."

- -

In fact, Broad believes that tracing naturally fits in with Workiva's existing logging and metrics systems. "This was the way we presented it internally, and also the way we designed our use," he says. "Our traces are logged in the exact same mechanism as our app metric and logging data, and they get pushed the exact same way. So we treat all that data exactly the same when it's being created and when it's being recorded. We have one internal library that we use for logging, telemetry, analytics and tracing."

- -{{< case-studies/quote author="Michael Davis, Software Engineer, Workiva" >}} -"Tracing has given us immediate, actionable insight into how to improve our [Workspaces] service. Through a combination of seeing where each call spends its time, as well as which calls are most often used, we were able to reduce our average response time by 95 percent (from 600ms to 30ms) in a single fix." -{{< /case-studies/quote >}} - -

For Workiva, OpenTracing has become an essential tool for zeroing in on optimizations and determining what's actually a micro-optimization by observing usage patterns. "On some projects we often assume what the customer is doing, and we optimize for these crazy scale cases that we hit 1 percent of the time," says Broad. "It's been really helpful to be able to say, 'OK, we're adding 100 milliseconds on every request that does X, and we only need to add that 100 milliseconds if it's the worst of the worst case, which only happens one out of a thousand requests or one out of a million requests."

- -

Unlike many other companies, Workiva also traces the client side. "For us, the user experience is important—it doesn't matter if the RPC takes 100 milliseconds if it still takes 5 seconds to do the rendering to show it in the browser," says Broad. "So for us, those client times are important. We trace it to see what parts of loading take a long time. We're in the middle of working on a definition of what is 'loaded.' Is it when you have it, or when it's rendered, or when you can interact with it? Those are things we're planning to use tracing for to keep an eye on and to better understand."

- -

That also requires adjusting for differences in external and internal clocks. "Before time correcting, it was horrible; our traces were more misleading than anything," says Broad. "So we decided that we would return a timestamp on the response headers, and then have the client reorient its time based on that—not change its internal clock but just calculate the offset on the response time to when the client got it. And if you end up in an impossible situation where a client RPC spans 210 milliseconds but the time on the response time is outside of that window, then we have to reorient that."

- -

Broad is excited about the impact OpenTracing has already had on the company, and is also looking ahead to what else the technology can enable. One possibility is using tracing to update documentation in real time. "Keeping documentation up to date with reality is a big challenge," he says. "Say, we just ran a trace simulation or we just ran a smoke test on this new deploy, and the architecture doesn't match the documentation. We can find whose responsibility it is and let them know and have them update it. That's one of the places I'd like to get in the future with tracing."

diff --git a/content/ko/case-studies/workiva/workiva_featured_logo.png b/content/ko/case-studies/workiva/workiva_featured_logo.png deleted file mode 100644 index 9998b471049e5e561af77feca573b2eb4c77e95d..0000000000000000000000000000000000000000 GIT binary patch literal 0 HcmV?d00001 literal 5980 zcmcIoXH-+$wnjt+1Zj~XA~EzsNkS*oP(oKa2m}ledJ;+?bOZ(IC{?H=iK{#+;`uf_x2cjueIl#-(24;WA8OqxXuGrYRWs5BqSu%>T1e* z#BrQ>2a=N#pZhw0pojwtUd05jkF&)SP;O`vxDCz<4OGXV?9h5>l#RD*H(G{-73IB~%$GYQDSZnm( zSciXMrT!xprsRf3;c;#TIGpp})zh)Z;c@QvI2WLj(x24=@@k{39k9QIe7_R>7hklS zgD2Vs;fBKi|I84~;oq=;ib*IzM5L79A_z%h1QJC0DhMSA9ImJY5mknYg8stV{F~_i z!YcodSTKa;`z-zT{e0;JU;C#8e~U(=ESf)y0*lk<`VJyHB6BWD(3 z?*9O=W(Qgr%*z!D>@4}rAnlhN@+JW|QPR)ZKf%4uZt?OE*2kseHEU=q|9EfoHPgKh zEC*wdcCA-x-@E*?vr4n#>JB+w2Y3(sUa7tQ-q5>_t9Z+fs>i)hF867E0ccYu4H-vP z55yZ;GJK^D<^3cO@`#M`OmWrO^6~V@O30%<3P|LO40FOz)Oo1~%>XMyNGg|v!X9?_ z?Y)hSgFbozC5Uhk0?@VM3j%-hJP{y`6A|NdI7;D6Fy+^{N_l2#c-Y3c!&Q2pz07oU z$GFR^gNqR-t4>)i`Zi4QGX3Lf{+-OVNnB=)-`=DBn*hGWFLU=f5=(Aukx-t&?OZ}> z`Sj)F3}MCdjpK_&ZpT_P_~K6aweM+ybBIl|65+X?xY!%9GvkWeMLoOJLHnJ#p9zP) zM{2=(lvL)FTnUw(Eqpm|W+vKz1EyK^VBI+CNp<q+C->B2Z%q{exD{h&2YFLr-y5yaE3wlb6Eq0v=>8aC-KelC2WI7fPxiM6NH z#Q475sxJ~F%|^VtypN@xts76tI!Kak{<^Dw6Sv6cDq;k{&A`a# zKGpm_wTr2044y8h=Z?|qZwxQ6DZZe{g9Ozjjxqr%$wr=j$?t1spU%up7he@)Y zdLv1QN@{#=y6$U>V!FKV9($2V^Oi1ts&NK3G%KWAJwGH?NRfF<-{B*nH)EE%kk`4N z0<=VAszndball4-A7Z~+rdS{fU00{KYO~ABwR|~rh2M-Go-B+>aQM=1XjO9Bc~2oy z-(qkkPlc+wTB!n_q!gOzzd_RFUgep(G_pDYo~{Y-)7cx*%pmBDr_5xlZy!EI*ty-x5r1daZd%Hz zm94qg4`AVzg~{oyzo8eY?+zFNu?B7&|4{f=;xS;gx#Pub7g<1hF6A^^)v`hp|AS^8 z3LZCX3Cvh(Ha5C8SF zgGKn7iU?8{D<8_;HO!JtJF?0Yc#^jj1U(d$=4bJiDE7VQ3TBlY~ErPAN zhO#qzXL`Qz%TFPMyhBHfAL*(4K+Prp2!A|vELhD4uBLG+B+XLM6j&uTEQ8eDQ>9l;oJoo@mn5@QUxR%xR6;ot(`rw>Wjx4u1QZDfIN}_QI zPwv7}fI_sAdxFoL%J}+`kQ{yZV>Qa4=CdWwu?9lbu26H4SXfmiyQ#2li}>@H9_92Rn2&W#dUcZPM{DiU2PHM1=K}{0Rj9~vm-lMf z?Y&PPGTDL%;{-s?O*%6n1fr3u_8RRqhQh$L9huJ=GrW}bvyPp*Ki_%yb?E2TSB7_N zVCD>bd1~_6(%L<7vZRY9(?(chsBV7k#US<3^&G94B9HyoJc80ScarayMEOe?!Zx}~ z!NYmpv+|M7CvUC{`)!)Zn|SW|e@j^5aE{$KxJd>a99ko{P^GGYnWS@dHRPCl9PMs?jQE?L%m$ZX5 z?c3)hEKX~R=Q0y}1+idR^*!ePmp%0@cl-+N(ynzn8l&{rjT##IJV{s=UNuqAcqUM- z6f-`vC{PQH)#;1oyxm@OlFB8*kcEsp-9Ykzwa4UBx%*edPQGH8X5|l%dECa;-|?M1 zL-&I#u5uOI<3X%1Zz~5HPu!y8Q*J&jM+UDV12o^r5{^xEzrHH4063`pwt4x-J%ScQ z8X@(u=rH!!O+OmqpT2!@rP9ylcU?07j#C!yRdSWdfa$rfmp+gGm>$P(lj_|S7B|8% z$S=InNbt)!pPlU=-1WCjaBSsZ*@@(RG?=f=rKQ~3-F=7=7`W?Ja(djqAwc-Vs582#@ z-T3H-7ow$&l&Hm&pQe{UG3ViCNb$ip5p}r%VQ4i?&veOlir~1*Wbq zNF+Y^ydR8Qgi;mmEWImXXFe7mlGYxQ_ssEUJ$mzrT>9MBTY2x9EoIt^aIy%B(+6S^ z3q36R@d0Lq7O5b%ZYHWeb)SK+xRd;Y{xzj4<)+*^yVc?N+cF9FzVa5hPY=yZfIcbn zuxrTa{Jv}b$unkXuI*Z3k1GI)S_lLwb^RWIO;4<-mp18X(80<-+cM1eo7srkx>WvN z%)K!_ld(s6$lq0*xaYjQ?Ogj+KF!}Nb%E)rjkT_D&_!m-am5;qWz2`l3uf{M4@|~= z$zERX?!0BBwbSi#kkk|2Gnp0xeY{nIc2I) z(V-;|8ykh`c4%QMY!EiNj|zFQOvK>uWz5RuAlO^sQq1mUV-N;J)D< zOsxL7lgRYF8_jx?eIM;;VJ25igJ!$f7OH+M_Nm_Re=E>bDVufJ5IAfw*G zn0AzaA4qs;CfDH^R$);O`*?HdvYNb@jJ1J_2=SM>g|knqJ=e^kNU z^=qbUJuqVm#VhR}+P8KHHDOxS=V~=pdb~Vh2wIjg1$Es9JpRkoAKuP^dN#Q_$yP}5 z2E&7LdZ9+`ObrR^L_Np2PS5D^vTd}wIHy#e*^9kJ(Ka7YfMPal<++ii`;R>I1_ungTnj$kf{dPu;^TIW$8--&~?| z^|ik_ewo;YTAjt;7mS!ikwGK*2~tgELV-^d8MDZ5R<1Dk#@jT zl_|{?X0A<~ZjLe4ma?0DPTp(SuTP<$Llryp$oX)Tp>^f4^XFW(TA3erkS`~s5c_nE zwHiV98co@MWWK*;py8l+d)9)I?Tgo$r4y28NiML9&zgpC==j*Z46R4D~%(CNagw+kU#`koFR^&6hQQe%cnCI#B^;m8foJ(#;ly#t~w?}p}TCNNS z-(6vw@yy+SHjPtJGF_4xKnm1p`+|=eW+#t-z(?30<*<+LXCOfu;)-J9qcwcRO(}A* z1DC?c4na_Q%{GGgot!e{N}|b`Yg(3567dAvrQD>}xZhLS_VNjBOLk$qK1*Y5S*EG; zu)@m^%#8!gNhoq|p{s1>UE)m^B4^>Hvr5bHOOhe+2eHPJE#WW;_N6k9OiuNP1@kQD z*I3$&F5jDZ^~#77-o;l}UO&v>P1CB{xBS9I&F4pZ zhoRNB53i29=4+C2^`m-NOm$61qvv@9rrVzhKoo5f#PpUCn+{^%M-MmB3!LTRBQ6nQ zr4>A+hTEdv=CwW7x*0hmpIhb+x)!ka&t!cwE+3qc)-k26Xp1)JABwoHyu$+wA2hpQ z^%bMios)aLKcxHSW7%bh(xC6Kc+C3yv|D-Is{Lt8YN!g&pEOAi)sj`fZ5;7Nj?P;QCd_`#jBY(48!JN{eMlCQ%lq|8GR7QQ(Jl(PZ)IIi`&IkWj!!2QjT24eJ~StC zvxYok+{v4?o($;*6E~4po#@ysMp(tFN;2jciFa%W&6^ZKJhgQO`AT?rty$u%hbaj$=Rn~ zk!w!=`yXU-OC*J!YbH^#W~Yy2G}$hBl;kFbhVvg~E)L6$+3gv)PwR?)PP?)DG3x{V z(Z!Q;M~^E6flsQ~*xPaR36ppVuyB@nuLS=>GJxhtuA{nOy%ua7>wo?sy*4V+vix3c z^>!wLz&F41a7dn;X|4Q5$U@X=>y@3y{O&<|c`aL*r8v#)Dpx7c_@LF*vFUe53%05y zfIGjVOJ%h$Su99y7R?>B&2Nl%)Fvn$_|$dF^aq#s!Xo%|BOj}zqMeOn%G90~Ipk@m zq6paMA;Ss3buI``Cqs{wG~$-~w0)XsdT5VmCWH~bU^%D5)QehAxOtBRN22>dT;BxK zq07o9pc_&et(Lru@(n}mRc{pCFq23!J<6!u{8_h~%<7acF>JRet?XLIE@kO2@!2rp z5C2MC;*ULQwfUt$HiI1%h0r(GYTgrqcayJMc347rzgMlq!?6bLGL|~Fvt{cfDrdO~ zgGO?eR~GjamAgBfcB~s+I%&R7cJ2Oj>M6@iolVY~*Q>>wwLMjx5eLAAu^cv-RKT5LaZTa(+c_fb0G=Cyz zEkB&P`1<{P7VG3cjt+?M{|aIa1EpS=`1|rcYISyQjB`G*B5=cHyb^rwW#L!~0wISr zQPnKcp^HhVI@8PsXfq9W7)^+Gu}L#X4vnz~9@ZEEzyRY~G6%ph2ZuKjvL&+@ zyZTKpGzx6{k;)}8(kdow>>)$AA9`f$Cm6hrCd#CHX0C+U@3am4=ijP>L=vf|ryp?-tj|c3c*Gpw~(~C+`8Z zCxC1UO_FvYF<6w54ImUAj|JT2hd1q?fKGm5>mSr59Khke2T5?uKPqfgd19NC--ION+vD z=KU4#eYi2_e!O$f%sDq!Uss)kke(0&1A|0UL&fl4UHlhe1bF{)u&Aj71_mRqrivmM zns<^P5N!N;d34Zcs(aZV^$EyYr})7Dgf3UD)3&bZ5HzE*q-is*4XHJ(AGdT%kLXzF zF>RwZnE7Z%<)yq_6T?nGph_K%qeQTr8I%$5l$E;e%o)x|ZnO7rzp9rZ}r`wb7(fI@2bjrf#ZGdjIs1I#Z1TXkkUKqwv<-MGUj}bNLJo{QoIvqE1OqA0`b> z!`cM;yq|wAL2|Ys)a%Ul$<6J+XJM5568`# zCn*sisbC({0?sH@8gak>5Uz`XuUQg+9gdvT|G$i%AMc zLmeAWffNb3Gl(~?z3V1o7izR>%xfNtU#VZJSe?qVpjX?NK9NICoR|8r2bM8cB*N=3 zN^G=B@@hpO&-Lh z&fwNj=eYT#vYL}n3@UH_!n+w^ShD=n!7R5%Vve6B6*jFiQCKe5-!76 zX}m)lfE)9!8z$REI6AVP$cVG4U)hWwyBB_fJZ0xuBiWDzUWoCw=(9h2h8)-&w)vt_ zcWntg`-@)?MsiG;wqHA++d>FrB_E2%AI9m$$@{DSPANc)0Q=@R!ij}V2>=&-MH4Ur zT`m5bQXfi69uMu?Lg4wN;48p~Hg{kH2CU^T;WdiVf!Wsp=nphDmDzs-SABr6r0Kio~dS%Ke;(MGhvRR=3n{2ANBc zA-lvGEJRhGNp1VYIP-@96&(%H0q@vAHpHTsnIIc%r4$ekNl)9cwn+(Q~+YnABp8!!f63f#%@6#Cq? zbxY!EMDjOjC}dlrPr^z$Xhy`X!8sX$q>~{&nx0ETI!Xg01BI*b+k;+hCaSMDSG#I0 z8HQ3MDyQj%3k{eUOiOp)bNmOeQ5`naNV)XvA=J?{Su}#=QQ}{B( z4fP9I;){lHm`koN`N1{ zS5-DPyOYM@EEKvSa2G5z5-J>`*c z%3u(wh*G*o1y$O;>m%P=m|zGQ0JRXl|E&Gu3M}39YT^4$inA?-f3B8H|ik1ZX3X8(xdzOB{IN%YOn55c~w@ zV{oZHUipzfkIl5LYvBQnCC-c@tg=~`{&iQ5c}JlUY2p-(6A( zg6HZ@t$HKhn;8m1a(>TAJ?+I0+L0W@y>eGLQcte}f=!3_-2ULLW7^S*(VzB_a&t4# zu>=)aStJ3a{N6bJ-Bx4|F-=#>?QH=?$V12;12%u_dYhZGBSR7O`ln+$~U| zpD>TOVw8FUYiN!feWUtBG1OL6g;Ux@dqa&H%tUqw8QiYItr<2FWrPRANH=#Uyf_*r zR$Kr^jrxRovF(!h(X7M8hB(^D)fYZV<`?q(U%mt%CBBwjA)0UjtUk&5_1;&ts#Y; zFrrrU`v*uW3!{T3cA#_ze(d#x4ryuP`zNHV5KJ2-essH1o{i%GGg@tbKKp($YD>Ey`kiaF#YDMNfbt}B+5wRwZ``OZ%83iNTbJ(V0 zWqA;=F?A{z-xa_MlK0cmib#j|-TsawHzBbJ=3n&%?AA0wj$DoLiIbyZDs6_d?%n#h zK5r>lm^xbgQp^mg^0n~B1Trbf$x(Tm^9Hwy+Ithnm@ReEUH;@we&OK~fcAc+bQVJB zKMfxbM#ApI?!Hjw8I7{09_e}4-O4MjN;CBt(pw5J7;Xi{jdop{Ae(j)ILSks@Cn~RnB(K&NWcPuNTI^OKY=3Sj!W=0N_YeXx^}VbzscQL! zSFDvnqB2J2mNGD1md9=MDEpHy(X-o%g)~cTF@DP!@i&+y*~M!(N?~bOdw&24aNMeBzuzhv2MMmi;$8EFiJBVYu!}dyZ>cF;Npf9rZD>W>*ExEgr{mEy zKdJ^dzL zK*}vu#y8wbw|*bV5+Sk)(2=Ha5|5Y04mhHcQhpG({b>*ay0yaZh`fK~j>(=@JkS;X zl=fNgWtBQJ;z)3$8-isNQc_v^@qWYj;;XdzYd#>zM zE;Du=ommj!{r6ed)(7B#Pn*_SwAPSvUwIGja{#P3)Lj}?uX9#yw%vse&!MKPZEgcH8Rd)B2TbHk_vGCQ-=1e zUvn;n_bU_)&tT49N;!(gx*vF1N|Xn^ViPznPlJ}?**$Yq4rs771rTLASV<|G(;tNt z5to(`n~vXr%y@7QD>kNRdNoa zK?8uwSYt5~6iHv!vA(7aJb_eEz?Pb*_&NlyOb~#qs2cdXIyo@m2e~hJU5cD3Buluq3_06Wwu^#;Dwaw z5ANM8+SX<$JequKRj(ZKY#18Y#d6FI)_|OhM{~cay;gk!bq(WV@u5=&<4S!^KHOJCauC?NoxzIJH!Th3Zq*l?+ z3G^2iAdXzj54^`e)+&xX3F$hDSqB?Vp5U1genx z83_Rz2a^JJy1Vc{`ePsF-nu>i(V;cI+c-fd2m4yhjb!kdAA7zDxwoI(OH_J1mZfr@ z$Acf)g|rbCgov=y4`Kz9Di=s}NaP1S;!Lh$K@YvtnjF4OpD=9HtPog-Nym>3(dCiM z7$i6lG0~tT>^<3b|0F?fa`jWxtzVE}lMmb~=M@&33XRO>!|QXL32T&W8Xx6(SfZ23Mcm}Tr}|j$Kid8NTQ#zIFEl5aHZPV4+zw4> z4i^7eJ<77gM`J`0xRWG@H-?SFCdtlAH%?S_Qos@9!gk*91Q_`=4SAw=Ev0|Fk;>rq zo~BP)2!dUTe?`Ho*U=&y@>M%JXtZK8xMWmH}H?; z!P{(2J$5>>xS#A({st;8g4+xD1(n@ba38eo4)#L(mFMXKYgjv;uX=HwE_1U{Y#o`- zw0W@=fTA^vUPa+v#ljy&0K(A>N0waMF+|Bik8l`Qa=NB0DkEHuC<>u zjmpc+4Q+kWKVD>UH3~&}H+y6Q;-2c6kIAM+doS$U3#wd>mFoK^Lq5}*{seS`Qg&&f zmj$0sR*b|=7CEl|qi)xw)Rv~n0~q$oMNvOr092SHnG zZPTIATF7+gBfsdZt+SG$eH*}$!ns3BHjVhqqAaz@;;rMPDXt_86(&Au z9l4s2i{Q6E%uCErU;F z*=ha-EDHdn!ftL(&Zv_J(PhLg{YO@-oy!7hYN2e$(e-Jy<~ff{<=x`I>rJysbf$AdFl z`Qr9M1_E9oPxh3jQ|3ASFcH$V#^R%Y?Jksf>|z(s!ui!v+3j=SL@GpvR<%6qo5W=B zP$yQ7a4^4zVV~CAbQx zT6a1&dIqmKbZ%&8Bj2=lJA=_Mp=1u;qA!L;1Z7D|%NR4ls@3M}Ws2U*`?)bxs|^*H z3){2i`HG41!6n<|MWvYo2tqj|UodGzD_;ePm$1%$Sr1Pq0MwIF6MW>bwE8itK+VhT z`KnA5+zzWEbm)J?th2%|39%VuI$B+Nc8r?yT5=ej@fBk2vs^Nud{k^x)vdv)|IuB9`kFQOI~Pvd_9Q_8FX3v zY?{VIM_F|Kp0ZF7k$npZZmT&Uz&AxDaZ8*B0HC4F3!Op0avi4YZ-6RL0vi|{Ygbj6 z(nw!Ao}ysW@Dzu`sG(^hm`Yl4+1Oq7Pf$EW_rbbDYhSu;6o(-f#KoA7`O>B3IEa4V zM%8(Im858SaUt5XdGjB};^(D|74#d1J>+|1LR2838{JFw9E@WPnX3NJk0Dr9V zPR>kMe%(gMa)kWKEro^V9Or934Zx!AKg=Cfyw;caqMou?R(&(4!YCEE+PoCz3DweP zI5e_}{U^c9E*c6umHU-_T`$P`E+#(Mh7dOPFVfeeEjn?VkBRPlnd;1mj8$tqLVVUE z0AoIbNvSmp(2)Uq6^{YW8~<<9hr(B1P+tyBaw~yYTdjWjn|sDGooRf>JmP?5A?=fA zi3*gNyxbv64KkurAHd(oUTC}H`&@SFo9w7}0Jf7Kq2f$wX|%n!)jcDlHSOL~@iMMn zF#nc!zQ4sYd3fMdwe%`>1QygLIynfVNW%S)w{A3aP8)yk#Lx#0F51Bf$4H9s{Nlfm zxAJJGH+Dbnd4kV7L_h2OrMp@BJ;3J^#4LOHGPoU3qs2;zUy2yo_Bo6tLYh60^mP(< z9D~)1KRcmd?n|v?;}h#ypqRDSBPZ6KU)EN$lb{#x%vuO*_|llwNmU<@`vNybwZ=sr zMe$8S8_;~RU5EB)AwiRiY*_b!<=u*Q)0;-E7d#AS;koluJGu&$EZt2^JELzimDQfH zHv$9xG?D%&{++s=^Ls2PQ_|UrV6N7%*w#dA4hlwr(a|eamv1n0Ok| zfg1wQ!$z+gRjE$vC?4Kns~&phJ0@Z|D$fP6)f^wZSDO2|`CbS0KtViNs6E~kp!9GB zY^A3f&2NPnmW}(Z*Erg1k!eXjn?Q*bB~emEdHHE&eV7P|JDX z2N|!Pg-_o~ho`{%l=I9boZxv)>@p`dn;@1zGW|;BzxSc&m5F3Y>shIHV-aa8NMoa$ zLt#;IKz0sJ_7CAT8$vhv;X5eRsX2Fzy$-gVCBF6S26yHVM-poF7UY0fB*T0`(&yol zHH^=9v{ubMRHAymUI_3ziwJ3X4;uS7sOJ4=e!f0oY1i61rfPDy#q&m^EhqQ8|vK+-v7yx z!LPneGyANgejy}v3SK=J;f1*>jqNtq>F4s@tl^qK%axMBx3F3hUyaeP#M=waDyKS6 z3P{fRcz?n7Ptjki<4_djCVg8qz(gp1xxZ2SX;Zhi=U^p6QiAjPLHsl}l+oUaYBH3e zgaX9S0THyo40uo|LlQlW5KFdG;VUSZbDDAeDsWp*i(`HAL2vgqp%P!(7nNfNy^NU} z+?{O`N*7-fV^6v8TamzdtTOB-GW3BgySB@HwI(L?pM=36m?h<(p zr9Q<$T|k=!ZuEuIw$>zQR8FH8cYe74Rykh+uziGMqf6L<`s$U%&YF0+#GmosZ>d&8 zU&y{7^Vu|8)rd?59-6hsL`LQ90~h87m0$BkaeBh`vEE2ta#AwI6i|mo3;%QDo?|lf zyqRB_bf2;%?v;zo6&lvjGq7a`(L`qE!DMeKn|`BYh|-5j*)Bx*ruE~$w=Jpcr?nqz zFLmJw$}n`A`a9N*+m4KG=DjEOrNcd1fbiQ|%b$(AR`Em%@@hHC>Dn{?I&euAA$Qd5 z{ZV?O5_cp6H?XSU_irKQN{r+oDs9R6IuQriZH4$oqU>LXz2k(lf^DLfj0i+{^_v;I z&r;!X7EuNk;n}7NZ%>2H7v4KjS#^hh9h75j^Jb{n3832xC(0-n{C};_i@E~WWfIQbbnCnh1cv1pV&bH*+v5H*Iw?i z*Gs(&1z{8^CGq5w=0(Kx>Ip{SldU$9SH947=luc%55EbB`qh+vE^Cwm zzXfLNPyIE)fO)9#Zfm2m>!*)v#Ky`~!bk*}DLeV*Vf{niKsKDYpXjH}cLV%?raX?V diff --git a/content/ko/case-studies/ygrene/index.html b/content/ko/case-studies/ygrene/index.html deleted file mode 100644 index 9afc1ec45b..0000000000 --- a/content/ko/case-studies/ygrene/index.html +++ /dev/null @@ -1,82 +0,0 @@ ---- -title: Ygrene Case Study -linkTitle: Ygrene -case_study_styles: true -cid: caseStudies -logo: ygrene_featured_logo.png -featured: true -weight: 48 -quote: > - We had to change some practices and code, and the way things were built, but we were able to get our main systems onto Kubernetes in a month or so, and then into production within two months. That's very fast for a finance company. - -new_case_study_styles: true -heading_background: /images/case-studies/ygrene/banner1.jpg -heading_title_logo: /images/ygrene_logo.png -subheading: > - Ygrene: Using Cloud Native to Bring Security and Scalability to the Finance Industry -case_study_details: - - Company: Ygrene - - Location: Petaluma, Calif. - - Industry: Clean energy financing ---- - -

Challenge

- -

A PACE (Property Assessed Clean Energy) financing company, Ygrene has funded more than $1 billion in loans since 2010. In order to approve and process those loans, "We have lots of data sources that are being aggregated, and we also have lots of systems that need to churn on that data," says Ygrene Development Manager Austin Adams. The company was utilizing massive servers, and "we just reached the limit of being able to scale them vertically. We had a really unstable system that became overwhelmed with requests just for doing background data processing in real time. The performance the users saw was very poor. We needed a solution that wouldn't require us to make huge refactors to the code base." As a finance company, Ygrene also needed to ensure that they were shipping their applications securely.

- -

Solution

- -

Moving from an Engine Yard platform and Amazon Elastic Beanstalk, the Ygrene team embraced cloud native technologies and practices: Kubernetes to help scale out vertically and distribute workloads, Notary to put in build-time controls and get trust on the Docker images being used with third-party dependencies, and Fluentd for "observing every part of our stack," all running on Amazon EC2 Spot.

- -

Impact

- -

Before, deployments typically took three to four hours, and two or three months' worth of work would be deployed at low-traffic times every week or two weeks. Now, they take five minutes for Kubernetes, and an hour for the overall deploy with smoke testing. And "we're able to deploy three or four times a week, with just one week's or two days' worth of work," Adams says. "We're deploying during the work week, in the daytime and without any downtime. We had to ask for business approval to take the systems down, even in the middle of the night, because people could be doing loans. Now we can deploy, ship code, and migrate databases, all without taking the system down. The company gets new features without worrying that some business will be lost or delayed." Additionally, by using the kops project, Ygrene can now run its Kubernetes clusters with AWS EC2 Spot, at a tenth of the previous cost. These cloud native technologies have "changed the game for scalability, observability, and security—we're adding new data sources that are very secure," says Adams. "Without Kubernetes, Notary, and Fluentd, we couldn't tell our investors and team members that we knew what was going on."

- -{{< case-studies/quote author="Austin Adams, Development Manager, Ygrene Energy Fund" >}} -"CNCF projects are helping Ygrene determine the security and observability standards for the entire PACE industry. We're an emerging finance industry, and without these projects, especially Kubernetes, we couldn't be the industry leader that we are today." -{{< /case-studies/quote >}} - -{{< case-studies/lead >}} -In less than a decade, Ygrene has funded more than $1 billion in loans for renewable energy projects. -{{< /case-studies/lead >}} - -

A PACE (Property Assessed Clean Energy) financing company, "We take the equity in a home or a commercial building, and use it to finance property improvements for anything that saves electricity, produces electricity, saves water, or reduces carbon emissions," says Development Manager Austin Adams.

- -

In order to approve those loans, the company processes an enormous amount of underwriting data. "We have tons of different points that we have to validate about the property, about the company, or about the person," Adams says. "So we have lots of data sources that are being aggregated, and we also have lots of systems that need to churn on that data in real time."

- -

By 2017, deployments and scalability had become pain points. The company was utilizing massive servers, and "we just reached the limit of being able to scale them vertically," he says. Migrating to AWS Elastic Beanstalk didn't solve the problem: "The Scala services needed a lot of data from the main Ruby on Rails services and from different vendors, so they were asking for information from our Ruby services at a rate that those services couldn't handle. We had lots of configuration misses with Elastic Beanstalk as well. It just came to a head, and we realized we had a really unstable system."

- -{{< case-studies/quote - image="/images/case-studies/ygrene/banner3.jpg" - author="Austin Adams, Development Manager, Ygrene Energy Fund" ->}} -"CNCF has been an amazing incubator for so many projects. Now we look at its webpage regularly to find out if there are any new, awesome, high-quality projects we can implement into our stack. It's actually become a hub for us for knowing what software we need to be looking at to make our systems more secure or more scalable." -{{< /case-studies/quote >}} - -

Adams along with the rest of the team set out to find a solution that would be transformational, but "wouldn't require us to make huge refactors to the code base," he says. And as a finance company, Ygrene needed security as much as scalability. They found the answer by embracing cloud native technologies: Kubernetes to help scale out vertically and distribute workloads, Notary to achieve reliable security at every level, and Fluentd for observability. "Kubernetes was where the community was going, and we wanted to be future proof," says Adams.

- -

With Kubernetes, the team was able to quickly containerize the Ygrene application with Docker. "We had to change some practices and code, and the way things were built," Adams says, "but we were able to get our main systems onto Kubernetes in a month or so, and then into production within two months. That's very fast for a finance company."

- -

How? Cloud native has "changed the game for scalability, observability, and security—we're adding new data sources that are very secure," says Adams. "Without Kubernetes, Notary, and Fluentd, we couldn't tell our investors and team members that we knew what was going on."

- -

Notary, in particular, "has been a godsend," says Adams. "We need to know that our attack surface on third-party dependencies is low, or at least managed. We use it as a trust system and we also use it as a separation, so production images are signed by Notary, but some development images we don't sign. That is to ensure that they can't get into the production cluster. We've been using it in the test cluster to feel more secure about our builds."

- -{{< case-studies/quote image="/images/case-studies/ygrene/banner4.jpg">}} -"We had to change some practices and code, and the way things were built," Adams says, "but we were able to get our main systems onto Kubernetes in a month or so, and then into production within two months. That's very fast for a finance company." -{{< /case-studies/quote >}} - -

By using the kops project, Ygrene was able to move from Elastic Beanstalk to running its Kubernetes clusters on AWS EC2 Spot, at a tenth of the previous cost. "In order to scale before, we would need to up our instance sizes, incurring high cost for low value," says Adams. "Now with Kubernetes and kops, we are able to scale horizontally on Spot with multiple instance groups."

- -

That also helped them mitigate the risk that comes with running in the public cloud. "We figured out, essentially, that if we're able to select instance classes using EC2 Spot that had an extremely low likelihood of interruption and zero history of interruption, and we're willing to pay a price high enough, that we could virtually get the same guarantee using Kubernetes because we have enough nodes," says Software Engineer Zach Arnold, who led the migration to Kubernetes. "Now that we've re-architected these pieces of the application to not live on the same server, we can push out to many different servers and have a more stable deployment."

- -

As a result, the team can now ship code any time of day. "That was risky because it could bring down your whole loan management software with it," says Arnold. "But we now can deploy safely and securely during the day."

- -{{< case-studies/quote >}} -"In order to scale before, we would need to up our instance sizes, incurring high cost for low value," says Adams. "Now with Kubernetes and kops, we are able to scale horizontally on Spot with multiple instance groups." -{{< /case-studies/quote >}} - -

Before, deployments typically took three to four hours, and two or three months' worth of work would be deployed at low-traffic times every week or two weeks. Now, they take five minutes for Kubernetes, and an hour for an overall deploy with smoke testing. And "we're able to deploy three or four times a week, with just one week's or two days' worth of work," Adams says. "We're deploying during the work week, in the daytime and without any downtime. We had to ask for business approval to take the systems down for 30 minutes to an hour, even in the middle of the night, because people could be doing loans. Now we can deploy, ship code, and migrate databases, all without taking the system down. The company gets new features without worrying that some business will be lost or delayed."

- -

Cloud native also affected how Ygrene's 50+ developers and contractors work. Adams and Arnold spent considerable time "teaching people to think distributed out of the box," says Arnold. "We ended up picking what we call the Four S's of Shipping: safely, securely, stably, and speedily." (For more on the security piece of it, see their article on their "continuous hacking" strategy.) As for the engineers, says Adams, "they have been able to advance as their software has advanced. I think that at the end of the day, the developers feel better about what they're doing, and they also feel more connected to the modern software development community."

- -

Looking ahead, Adams is excited to explore more CNCF projects, including SPIFFE and SPIRE. "CNCF has been an amazing incubator for so many projects," he says. "Now we look at its webpage regularly to find out if there are any new, awesome, high-quality projects we can implement into our stack. It's actually become a hub for us for knowing what software we need to be looking at to make our systems more secure or more scalable."

diff --git a/content/ko/case-studies/ygrene/ygrene_featured_logo.png b/content/ko/case-studies/ygrene/ygrene_featured_logo.png deleted file mode 100644 index d0d69114784c8d418e8c0958229d301b3fe00505..0000000000000000000000000000000000000000 GIT binary patch literal 0 HcmV?d00001 literal 11569 zcmbt)bx@pLlP}KT9-QD3Y;bqi;2vOr0fyiVF2P*_!QDL&f(LgC0RjYfcXwOz?)UAz zf9!76tvgjS&+~NkuTP)uo-^G&9igTwhk;6h3IhX!p&&1<@p_$keSZSLy?$P(b%I_m z#84SMsHTG@)E(puhLNyvFauL6*nzCT8eouxr_&Hv7zPHO5u&9B)l*RxG@di+MPL%s+ymlxAQnh`k8)NlOPcCB#C6T9-!!sNyIEwuZ=iIfFI5RJF{# zKAHYFD3VR5?3a|r1L6jbLw)QT99wOBL$Q69O|4U}4ru>Hm`cZ`XKb6u`QKOV{ za0XNIu<^2*b8rJG`2^TFc!AtJ94wTaKn^~3AP+l`mlen%2;>*!c5i~4mnXA4U~ z4Qbhbt@ZjOLTwF&ItsG0ySuxyxpT2OI9suE2nYzU13B3_IayyhSY15rp&$=ddl#C2 zOOOV;m^(upp%4dq%D)mpW)7}U5$ab>|8om=jw&kuCD`8OUxs=$8M_C_k)4AL$Zlu% zcVGW-yFfL-|3{7gmfJ4EGJM#fM1%IQ$T`;`+vj=N;!i;PzPr%2M61Kdr!^U0qWpl?chi$CH0TDC>d2i z<`Db8M5ez6{+GUBXNVivLe|;Aj`AN12}1rGEr0^D-24)<(gGZ^z`t5Z$N(ky`6Q%) z0#ZCu(o&q%|H4}Qcf0>jtjzz2Wq(zM{jYofFL(c&^cq8dZ~rTJUmyM}{K59GG4K2u z+{5NM_%JX`ISSGeS|0O_J*u3-&={w{ScxPlfOGH{>fWCo$Uk30nbn*6kOcbK%pa!Z`bWo$D@>BlnzA5!$y&$!lea-$YZ0%|3gv>NBBz$mZbfM zgikFI_)2=^`%8M|`%L*t`rn`lZ!15;KMJdMf3pqW3IzRN!oK6hE{h$H1uWs_W$@q9 zu%{+|lUJaTu`DeN+S1C?5CjZt@>g|^r#34il77RJwEolcjpF@jhR%zd!!G}5lU^(= zB0ow@k7*|`_J!aXORcAG3^%q9g*f7)Ya|NNG{HMaNc2y7N1lD&;BE>fMjH$nc0nIQ zvSk%+!qxat5H4m92b0V$BQ^X07#~Cl@F(o&dYNhE;RMuNs?r z8Z0ybASmF_Pk91DmNvyogqk{rp7QuSQZ0kToe|i+;L^wV6dH&sGO@?%$W^abnF|f4lWh2$e4#!(j z#G!?&DQANkH(E^{{N^YQU)Y_4@rhydRzVUn!8XlY zOsrk&q#PiKn+H^1?1IroOM03gMu_l075Mi1{P-CB*Ef$rPIm4z>CXE5c0#PU>%XFkwHx zfy;_q)V+)>J2jdrdwIzRD)UX%KTmIeVCa$u2%TTLr%vgvrM7l3V}s{OJ~;5Sti(CD zM^<(aCDtOX=E=9e$9KjJ1&JUOcLO1hGhNnKd;K{f1uCh*!{f%IeBw9qosv0>4)lhcnj{SpidyW|`W=vo0x93^ zmBxougLk*51&!^`DkYx0(hN5NlRiz``?gOiY4fB6S~cRISw@?H(K)C7xddOZS9HPo z(Jk|5;Y)2U%4-8QbJkNWDXgenj5GCBm9DL&+NyQdDZ_)dOz}mDZ=`%W>oaU)&FQ+0 zGX3QPdtHhCoQ$J~(Qu}557%Ye{Sd;gA~tDF>z#NU-XhIHptgFi=K!7snnqFA@fXJ@ zctz*jnR-%kQmIn^gu7tuDl?^48=xPuFN%udbln`)3|o zsLcjn6Grf{-^~1(N=!-_%MePTbPT-+^g^g2Nq^$Q<;xds7gh(vt~NFdkS~uSR3W(h zWRT;hZ=gmxNM(kVwGhy`P446pKer zKxL9vLGPJASMl}IjyZY!a7fGRa+;orjyu6LDq4663?tukyXYbt>PNVPsot~2^95g%z*BRubO z=7bNP;$3X!08^7GhS6l;Lyc}+kU@KyK7wWHMwj?VsZR_vJaHH<=RF-Oog;?cs`#cP zqrnF1p*u1rrxDRQE_wk1YA?d~!T|!i&tDKB6v*m>FTz(X4Gp}IcleU6oL*Cd$P`U) zD_!;%q{7k{Pm=VrW$@mm5D@-WHUv*fNx2v&;m*A|n&!O`BujxG3)$;;tjlJ{nJM4g zve^R`Rcg_YnyVRuCX7g<|I{P8ao0Tsfu6A2P^0A)nsyNvPMDpa2@i~7L>{hrd3Ye) z>50(Wv0w z=p0=X1)#Vgw>v#_z*-NNlC_!TtBdHCq@o_1hkAixHpc9pKdjT^!lkqXB0sHh*~|^9 zhL6{RF9=sLi&q`dgG%p`!D1X4qkH5gwO*j8hcD~VktZBkv4!jfcmN>-;g3|^*3eJl zTx_U(ghbD1yR3e6X#QA%)?{Jwa9DDrhKnDWGbltBJQ^7;V*=l?j@>KO+@-Lf1)#Am z_SJicDiRlqZ?>54sJ4t|zF-uZ6}}Vzp{U`1$1K;KMFMvZ~?V;7G=RcL`0~-mIk1TX5I? z8p!#@SM*z7fVk;>G$bu8ZE0$;G;?8pg?B-JK~4|;C6uz^!7AXPh6nB7x@BkrDRuh?{XI5(_WqRNaPi9n422xQe^N>WN@ zDP}CsNM|mUOvV_mM7z5AB#a9Avxk-iYc|QufZ}8yy+4=`%|OX*DI^%q6tBE)^Zs-e z@qtUcs3^Il(^S$CFYgHDF6Bn&tIj)orVX>4G4~CO87l)8ChkH=6QOCkbdE++_+~4<<0m|rYsQ*+wtkLRK3=6G+HGe zhVkp4&_93v@RrmzmoX8cYU1mwshN%+xC+{VKzRnEg!WpaC=>NHO>)#fr6;-Z&=I#d z)3n}&G=zq}DY5F^=|*H~+3XHMV@p|>*e@+B3yB(tCFb&bY)YJ?_E0l7$J^b83W^vz zXliQGlI0a@71K>xb6WEl1Wl-Y&dH&l>k;KV_J*;1z;aPbz7LGzY5@?yscE|Io*J>5N4v&pw ziFp5-fTsaXS-Opce%%9pORro#E78wdwiVVc~d?lP~Pp`_3^@|vvnCF2EUHC z*eEtp`>^D2-2|vbIu%STEa_a{I{NyM;r>9wnd+CtjVsvbeBl9^T9@xTqgiOfjSRJ| zCP%ZdJaE08LE>-66@%G@abyV>_0&F#Idun@X_i1sQYy;J+n=`wL-sFJQ$|j(dJVK~ z9UVtZbN4TQNCYVsD`yKst!0L5=5rX)buSj?=^o?Bnb^Ns`;mbSK9YaxyNy6)A;;z* zHkvy$&T?oKoAD4LZT!AATeIQU!7d;~12u+r#U;Oa^Clesc5dgu{df`4#4NCvMClRz zZ7ape(3g@95fk65Bwbeh@8P`|uC7778)d?5p}zH*00Q1R~ghu zAH^8uH8owp;H^Rhyc9P5O09Qanm(6R-ZvB%hnjQ-Yyf#-B0>=F3PLtr_D9%#pUv$wkzaA;_0*Y%?MSmmlvW3G01co_F6ihX7hXp{q`%F23-KRJXeyt$ zvTt~9P(H%oK8U?q**TaQNayVOv9+@^mNP(uRJxVrz8pZs%vPb0Hkn*k-43$eE{GGO zmtP+aL32BvN8AdRh_l}4JdWbU^n=T8HLlj9d<-rj6LGVh{gwxV2}m;byCc~3-;9uk zjdbdtu>W*)Qmva)jigtY=^D*MH?g@<%w`^SUiqff2X;Ljqj8Y1_HBrC?KxV2$ zdO|Vat3r?^q9X7!&WYD=a*@8BWH_`JP1wlADT+v}IydKXGvZN9HP5HKLtDkaATsP& z)X_A&nao)R4-MY3F&@HZh3?6su+aeDjs7xImQ1n|HP>Oc5HJ{2D)Hv-2_;*;qP0)_ z()|lo74Fi!p4V;rGpi903kCsb&;go!h7ZeaOQ^O$#L*G(S6CMluMN=t&~p7{Owt6!NRzj4#Gvz}ZrK;DP6Z^r*JZ&pY zauOcD%4*iX$&Ko&mZQ9{Hg?Tty_&bhZZdh4-$C!&=Akk~e=m`#)vDI&q}}!4k3=J# zQ9tS*RIe#(Zp-f3pMx;j@Q0B#0G|%Op3sY(F?X>JcssU~p)T3#yVW2y{!ExIy%e;q zi*$anSVH65sJ7|&Q$P((Jqbo>VpPg zM^Tvh&@r>GYYe7vegsd+iLjizST!ThzR0-!;*{8rJzSqE%f71U@ynsXbzzidB|%{8 zVO>ru5w1F8kdF1*yi~mEAAV9t#l$Iv*L8&f*gdPB*|N-h>=s7YUPMpZgDZKmQ|J0= zSqNLFw$LB?2n~;TTNKL_>0x#Ua+X~g>fjK%9a_)c`%`jz*&0OxLrfS{Gl6Ek;fIwh z%|=Kfuull8YlMa!=tyuH0h!V^hxU`7r5VZKOukiB9VbZi-r7br)fx6uN*l11XujK; z__iH{WoYXL;9p09ha->b6ZXD7%Zj#Z(W|!fSVN(zP1fVS?LaRs3cGOoB@uKywydIo}?X!Yry^g1O zEi0{UvQ*4ue1lM*QITs-sh)v-4Lfs{s--@Xz)Zg6^mAp+`wy@^IJ)0ktme%mae=pOo^^Sv7fZn`-n zS3G>@vYe?ov?bU^u4BGNitH0g4<9}%7!3ZE?fIIPBDBxfXtJG*Pnu=B&Aqn|4amT? z2tvZ}+U-KlzF#`=mJUX#gwI0I>)1}G(=1U&nw6-Il8I2IU_^;V(=ja28Gyo{6LDGq z5XfYMsI<*-&HJN`{T{x)cx;T`GEv_yTh_i6yu7M|X!h;9786Z*^_?#it7N6G4s}(F z7v*l^bv(xq2@bq4kwkR*voH=4eh7K}THzP$rNXI=R#sNZQFtJaWue?zG zfHfn8IhiNN1oTgat@*AJ4wv~PM}0F2KN=$O3P&W+atif=MC=Wz-@U~gxi|FuGf8v4 zLY#hKlePtc+mMp1nEqitplbSzutSoLLbA2&eT+jh@LKllo><%vTjSodsBB%FI}@@Q zjUOC#09FW?liBrK8Tt4KFQY}2eu80WC`6rUDU>B?tBihFI4J`9Eq)h9sMRCzVfr^& zjTsZArOd}E31HxbOz`;LU;g^-n!D4)VNN|E?^Ao4!_nGYBr}CKcz!rnI~3|Y_dP>X z5=am9^B>%p-mb+!?fc=Hb9LOAMvzi zy01LdGFvLY2i|aqABs6*dqVNeH8Y7~z?rUt1Nm zjg6P1_*J9o!ps$7}AvSESOwJMucAWJ%ms* zWq8{Zzf=F4XokGcolZF`4E8x9zT3*(gUCspM?vYF9~2}&?bML`w-jFcmub&eXargj z2n?NMa>RxvDwN%+&$|Wn{HnbJtsKZ(^B=1?NHA)O9qdk0F}cDF>=Yvv?I~BM%m-ro z9>kZLl~(8VA;n%I6z@M)a6Q8hJK2P4R@%RU;<%S}&QB!eDF* z)!dw@6UDG8zx>uPi^)Yu8QHq>zzuyfmEYbDdblR<ZhQI0UOfE9#SL(O{FG6tlT_%ifb3d$r_mR9}gC{_49)6x#wY%>S0OP+6n>qT5M z`^skX^P$Ln^R~PoRZb3RewS@@M~iy~=hgzr!t|<5DGmjO2ptZr93+OLJI=~4j9;+i z4zK@&1Mkvuulu41nVFeeoD=W3bZZl`xHuGmpgj7rp_?JiE7k=zKK<1SH9`yRcm zKAn5|o<;D%m`T?6H3uKl=HJ-{N~Zm6`p{1|{>QnQ<;enxXls*c)&5;_WgHMo^+S^$ zjVc#?5K2E%Fx*lNOj*ugWHw#SC)Ho1;aq8LYKDj3#XDCz{HS8%-;A26O3GOyk9yHQ zv7cce_@rTvMi7oqSl?`X&z5DNJ^-?0Y{Q8w-zOS-?O7n0nR4_MLWj*lh|i7LEO8tz zksq_ppw-iFWC05ujKNVnksJtB7OA1mf0&M4B63HNM zi$whZw8<69pHY4m-Qw@~z~T)IcM^!T?dmtxi_R<(BRDI4UOVhnU_9 z(04(ktJ-%>)2lB!ps+`xk4`R*j9}-`9Ek%THLJQRlz=;WMyMg9IOR6aBg5y@B;#r?iX%OR!$uG?=3)%W4Kyl{iGQTnf{DW*PmM9syaU?i_vv$#e~wI(Vh_cjK{Swg{y{RJ}} zb_(DJ75XpX-LdLu+J?IB=bL*GcmwWk1xxlXyBqI+vn)Td!S!L18%)g366G|79t*o4 zlkqDor7gJ&W{Kv+nVe7l`jm<=dEhEpIef)KpeT(DKX0ceBNH@DBF4Gq;_C~4J13ac zKx)(S=qL849Up$9EOBTM_}W+SlcN&;m7AHT9({=rX95*=P#dKMj&|m%l}y872bix1 zfuC6f%b=skCsG%xsX(f^0%TO5y`rRX<0TXQSxx*{axTk*KkOSDC{#}$5p}8VyWY6a#M-1-hO;oF@g9~C8;OqT5EoaMaA2`3VcqPMlB3eZSY5~1SQI?G>HBa!@IHUg;&<`( z_RWfk%jO_XQ>W>84(s~jsj?dFnx{fL<@1a4>h}9#3{u|m z4INp@gXYC!EYZi^+X%eAzP{vWbTz|8!`F5i0{*UJUBAOU4#TnKXlp)U!>v%B8a0bO zBe#g!(XBb7fuY@lul_a{{KfX+;$o(p-}Csz*bA{^aw`!_zuOr-kx%v8r`>c3Iy&~R zj~~VQ7{}>mSUxzf_bTLqh%nE(@NPX{3ojAx23*mokAb%;3_bvA1vMm_!imz?3gZ45 z$t062h}{@^)xh4!5N*Rj{@MvapL04SEzq6apMP4 z4q#dO&axE4BG-uM#@qDV*|+CVX#Gy;`3aGgIEil*27YtMf4w+ zz69hn?pSTWzT-rOQr*U`+em7p21U)hxyKa_ll%1SlBLQ(EIwgV|6cU2 zNsB(HCjA?g((Ha`1$jhp*J>np8VrvbY5DbO5Ecn~TTb_FtRJqi?LA%3z1{KHS33vX zCQeP4pv0g1>N;-(={=n#5Bs=BVC7vBg%nSO=Y8KsBL}Zq@Yv5)^K037 z8mJOL7e3=4-rb3IV2MI^Ns~<)3#W@mU?oP^-rnkIYwzrT)rnicSXM|Z25Xc_mT_i8 z(*-HQRC3709MBS1=~M*s#vsA@O|17ac!Zeqd42Yw5c6AH?Fza%TxxMWNLjt^y9jG4 z%G{u>iH>fXn$DdGc;9T=<@04IF>zs|q~l9{AWYzq4LGHVj+?6O9j@Vb6`1#zd0`a( zYLJWMudJ-2DajXr1Yt|*2lz}8uc=Z*wCy4j2O(_6O(9%etRcTMqY=F0RleiZHMHxJ zC{Dk;!r>n&KJhE`@V>=slqu@P_PxE{x}s-Chv#Aq z4R9Pe1<@?p6B4@v)VLK%F+#I{#RNLIPbZ-vr(jf-wK?8j5<6D;_=zFLR2 z2UoZ7nrlZ!qJFgb_&!Gz4SMV>z!4od5JB2Tm5f}!g(~Bw^Tt|RM~C;dEtiv-Hf41p z7t0xI^Ke4G9FrQHpmY)WvjfOz_2{^ajpqMwcyVP4H5&rO$K)#cUCn;F7-{sFznJ6> z+Bmab4R%S=YRkUI`|fWH5Ey&y3(3}|as^P{wJ8n#t3;I7@6E=LQ>9N+k{Xx z#P#^LsPZxYsTpwOc+2Z8#h_fSFL1vlf~qfw2Ur(<+MG1yC+k?impUNRxi5rFWmlIAX8k!udS+}ts+Zy@1Zc?K_>M0-! z6B&2&dhP^sPd#oDvvIVdu?=?t;Ygh$>UHb<2x- z_Oq(z+YBY^>L=QUf3bU|I!qW>2H_2;|78yc{S-bi3qr<2#Ew7Rc z?mR&Fl2WUta-rzNtqv~%!4Y`Cw*&7p85tROZ6Trv6Ao=ps*;hgs4xrMKDR|Pl3o5~ z5Z{2c1(KBbv`6DTOyo@`%$MZ*23_XwCZPu0-a0k5jJhwAx5e&A?lbl6)U8<9nnR)Z z8E@m{a80|npXw&=8WeeF-wNnj_(@l%rk$Qg>8$ksXwWUPKnh?+1`EsJ^z)fvr8plXhB62v%Z^n?H z9l`N;x@T5!NYsQC>2HZGJe8SqC@e{A40V$Yv!mEy9rdRlDYHaM+ojnp zQHt|1=&J2TXTi2181i5W{AN8-j}%nZmf_oF{h@g)4i&=G;ZCN@5aZn;X0(Nt1Bd^+ zGR)Ek1>{aUA80Q9Iz`}56H?%5fl|PVU#2-YJi?jda^I-S^K>)Yt@8TkXc(7!l>5== z!xRgPWtjm@%$0lFrK2;dT!Yt+;sc}x5_PH zFGLF3cKU<=bngS`hfH@3$`nPhozBhz!Q7YJ{56om#tvYj|W#)okvX(xz}~ zc5-@zwdLa_?^}Ovr!9GaS!^@8aY}be@#i=21St0#3OwETTE>h9XkXv2A*CzC=QB7F zz@U|XahCP(1alv4VpaH;U`NzGzfO6O%_v4F0|^oP!-#oP$Fl`~7*-fiz=y*75NW&o z*vMZ?Ra7M?JjbBzKBy?WMxsiH5gL(k9gbKYS5J3p$Z=0HLU=JFCBl-u%8ZJ&x`@Gg z{-!p&%>2XX=@ZyG*+Xg|D1N`8lkL+kOV1DTWtGQ+q`Q<0#^A}_*}`ZFv7RLJCJ`lX zt`Qk|BN~IZ3AD#;JI6`PFJA5Ul84`j)cu+6=-w@~2cAg(jF1b25ywj)@#|Qz;1+#6 zl{$-h@z{}-qh4ZneW>6u5k$SSw zfKpxI=Y9pXauhpso>24%g93|-7f_)YtEGC2WsCQi^ckvG3<45Syms4kUmK7==M=-0 zdM{RKkAx;+ifif|C8t>I0a!>B-A|`u_RG26ZrWqqipfs*5`WM)ix}3vV2@$_dr|6N6G2+=GU6psdB6IPW z3Q^YsIy)IQcc;gF<*fk$Bt0&L2uphefo4XLDP4}}8>4@g_3U2YZ+E5l9V77^5)&-f z^sEoVqPlM<&|kREQLqEYqr=u(+a0H69%|QV)gPe5Gs}XhOBV`21Nc0$rUc8Y^;9lQ zai7(Ihc3d!$37h=64B%qoUc&lZA2gG=Um6gA6v;e%4x!LrsE826q4O)v3`g$M1mNC zbd?Dt_?CxkFVDg~SqgCgB^1r}v*llc@IkAoVsso%mTO zF`z|(FLV#76Scz#R_WK(H}scs-X=99S5Ols>0o^%hf4OeqmbU1&dYV;IP3-TBAyUZ zs}d|8N`TgIKE77osr{&?Y1!-GgQBN^L@AzO@R&%5O8Ffj`B=}O0wUBJPc9f@@Ay1f znjHiB{z8QB2F=+5r;YLQub_sG1ym^EdI1h7EkOKldv2XE-1n2X^mt)X=vvxV0hlCO z8fu?EZFM8My1GsivT^h9hDd@;9)8(>6CZ-w+U3T7j7YefEX1#CcQ8c=#$E8}*cLec#@dwMtz+3K+f5(rlP;CSCAR zJL&=N^yD1fozCM2!ikDyzzimbX7csDHR|pLwk1!q4j-t3K+7-wOue;{@o=OWk5=AG wLADg-pP|G5$2r9Rj}gXyLmyYSUQmEAkl$A5`U*}qe}|YAWK^ZABuxYV0}_K~sQ>@~ diff --git a/content/ko/case-studies/zalando/index.html b/content/ko/case-studies/zalando/index.html deleted file mode 100644 index 23363da401..0000000000 --- a/content/ko/case-studies/zalando/index.html +++ /dev/null @@ -1,83 +0,0 @@ ---- -title: Zalando Case Study -case_study_styles: true -cid: caseStudies - -new_case_study_styles: true -heading_background: /images/case-studies/zalando/banner1.jpg -heading_title_logo: /images/zalando_logo.png -subheading: > - Europe's Leading Online Fashion Platform Gets Radical with Cloud Native -case_study_details: - - Company: Zalando - - Location: Berlin, Germany - - Industry: Online Fashion ---- - -

Challenge

- -

Zalando, Europe's leading online fashion platform, has experienced exponential growth since it was founded in 2008. In 2015, with plans to further expand its original e-commerce site to include new services and products, Zalando embarked on a radical transformation resulting in autonomous self-organizing teams. This change requires an infrastructure that could scale with the growth of the engineering organization. Zalando's technology department began rewriting its applications to be cloud-ready and started moving its infrastructure from on-premise data centers to the cloud. While orchestration wasn't immediately considered, as teams migrated to Amazon Web Services (AWS): "We saw the pain teams were having with infrastructure and Cloud Formation on AWS," says Henning Jacobs, Head of Developer Productivity. "There's still too much operational overhead for the teams and compliance. " To provide better support, cluster management was brought into play.

- -

Solution

- -

The company now runs its Docker containers on AWS using Kubernetes orchestration.

- -

Impact

- -

With the old infrastructure "it was difficult to properly embrace new technologies, and DevOps teams were considered to be a bottleneck," says Jacobs. "Now, with this cloud infrastructure, they have this packaging format, which can contain anything that runs on the Linux kernel. This makes a lot of people pretty happy. The engineers love autonomy."

- -{{< case-studies/quote author="Henning Jacobs, Head of Developer Productivity at Zalando" >}} -"We envision all Zalando delivery teams running their containerized applications on a state-of-the-art, reliable and scalable cluster infrastructure provided by Kubernetes." -{{< /case-studies/quote >}} - -{{< case-studies/lead >}} -When Henning Jacobs arrived at Zalando in 2010, the company was just two years old with 180 employees running an online store for European shoppers to buy fashion items. -{{< /case-studies/lead >}} - -

"It started as a PHP e-commerce site which was easy to get started with, but was not scaling with the business' needs" says Jacobs, Head of Developer Productivity at Zalando.

- -

At that time, the company began expanding beyond its German origins into other European markets. Fast-forward to today and Zalando now has more than 14,000 employees, 3.6 billion Euro in revenue for 2016 and operates across 15 countries. "With growth in all dimensions, and constant scaling, it has been a once-in-a-lifetime experience," he says.

- -

Not to mention a unique opportunity for an infrastructure specialist like Jacobs. Just after he joined, the company began rewriting all their applications in-house. "That was generally our strategy," he says. "For example, we started with our own logistics warehouses but at first you don't know how to do logistics software, so you have some vendor software. And then we replaced it with our own because with off-the-shelf software you're not competitive. You need to optimize these processes based on your specific business needs."

- -

In parallel to rewriting their applications, Zalando had set a goal of expanding beyond basic e-commerce to a platform offering multi-tenancy, a dramatic increase in assortments and styles, same-day delivery and even your own personal online stylist.

- -

The need to scale ultimately led the company on a cloud-native journey. As did its embrace of a microservices-based software architecture that gives engineering teams more autonomy and ownership of projects. "This move to the cloud was necessary because in the data center you couldn't have autonomous teams. You have the same infrastructure and it was very homogeneous, so you could only run your Java or Python app," Jacobs says.

- -{{< case-studies/quote image="/images/case-studies/zalando/banner3.jpg" >}} -"This move to the cloud was necessary because in the data center you couldn't have autonomous teams. You have the same infrastructure and it was very homogeneous, so you could only run your Java or Python app." -{{< /case-studies/quote >}} - -

Zalando began moving its infrastructure from two on-premise data centers to the cloud, requiring the migration of older applications for cloud-readiness. "We decided to have a clean break," says Jacobs. "Our Amazon Web Services infrastructure was set up like so: Every team has its own AWS account, which is completely isolated, meaning there's no 'lift and shift.' You basically have to rewrite your application to make it cloud-ready even down to the persistence layer. We bravely went back to the drawing board and redid everything, first choosing Docker as a common containerization, then building the infrastructure from there."

- -

The company decided to hold off on orchestration at the beginning, but as teams were migrated to AWS, "we saw the pain teams were having with infrastructure and cloud formation on AWS," says Jacobs.

- -

Zalandos 200+ autonomous engineering teams decide what technologies to use and could operate their own applications using their own AWS accounts. This setup proved to be a compliance challenge. Even with strict rules-of-play and automated compliance checks in place, engineering teams and IT-compliance were overburdened addressing compliance issues. "Violations appear for non-compliant behavior, which we detect when scanning the cloud infrastructure," says Jacobs. "Everything is possible and nothing enforced, so you have to live with violations (and resolve them) instead of preventing the error in the first place. This means overhead for teams—and overhead for compliance and operations. It also takes time to spin up new EC2 instances on AWS, which affects our deployment velocity."

- -

The team realized they needed to "leverage the value you get from cluster management," says Jacobs. When they first looked at Platform as a Service (PaaS) options in 2015, the market was fragmented; but "now there seems to be a clear winner. It seemed like a good bet to go with Kubernetes."

- -

The transition to Kubernetes started in 2016 during Zalando's Hack Week where participants deployed their projects to a Kubernetes cluster. From there 60 members of the tech infrastructure department were on-boarded - and then engineering teams were brought on one at a time. "We always start by talking with them and make sure everyone's expectations are clear," says Jacobs. "Then we conduct some Kubernetes training, which is mostly training for our CI/CD setup, because the user interface for our users is primarily through the CI/CD system. But they have to know fundamental Kubernetes concepts and the API. This is followed by a weekly sync with each team to check their progress. Once they have something in production, we want to see if everything is fine on top of what we can improve."

- -{{< case-studies/quote image="/images/case-studies/zalando/banner4.jpg" >}} -Once Zalando began migrating applications to Kubernetes, the results were immediate. "Kubernetes is a cornerstone for our seamless end-to-end developer experience. We are able to ship ideas to production using a single consistent and declarative API," says Jacobs. -{{< /case-studies/quote >}} - -

At the moment, Zalando is running an initial 40 Kubernetes clusters with plans to scale for the foreseeable future. Once Zalando began migrating applications to Kubernetes, the results were immediate. "Kubernetes is a cornerstone for our seamless end-to-end developer experience. We are able to ship ideas to production using a single consistent and declarative API," says Jacobs. "The self-healing infrastructure provides a frictionless experience with higher-level abstractions built upon low-level best practices. We envision all Zalando delivery teams will run their containerized applications on a state-of-the-art reliable and scalable cluster infrastructure provided by Kubernetes."

- -

With the old on-premise infrastructure "it was difficult to properly embrace new technologies, and DevOps teams were considered to be a bottleneck," says Jacobs. "Now, with this cloud infrastructure, they have this packaging format, which can contain anything that runs in the Linux kernel. This makes a lot of people pretty happy. The engineers love the autonomy."

- -

There were a few challenges in Zalando's Kubernetes implementation. "We are a team of seven people providing clusters to different engineering teams, and our goal is to provide a rock-solid experience for all of them," says Jacobs. "We don't want pet clusters. We don't want to have to understand what workload they have; it should just work out of the box. With that in mind, cluster autoscaling is important. There are many different ways of doing cluster management, and this is not part of the core. So we created two components to provision clusters, have a registry for clusters, and to manage the whole cluster life cycle."

- -

Jacobs's team also worked to improve the Kubernetes-AWS integration. "Thus you're very restricted. You need infrastructure to scale each autonomous team's idea." Plus, "there are still a lot of best practices missing," says Jacobs. The team, for example, recently solved a pod security policy issue. "There was already a concept in Kubernetes but it wasn't documented, so it was kind of tricky," he says. The large Kubernetes community was a big help to resolve the issue. To help other companies start down the same path, Jacobs compiled his team's learnings in a document called Running Kubernetes in Production.

- -{{< case-studies/quote >}} -"The Kubernetes API allows us to run applications in a cloud provider-agnostic way, which gives us the freedom to revisit IaaS providers in the coming years... We expect the Kubernetes API to be the global standard for PaaS infrastructure and are excited about the continued journey." -{{< /case-studies/quote >}} - -

In the end, Kubernetes made it possible for Zalando to introduce and maintain the new products the company envisioned to grow its platform. "The fashion advice product used Scala, and there were struggles to make this possible with our former infrastructure," says Jacobs. "It was a workaround, and that team needed more and more support from the platform team, just because they used different technologies. Now with Kubernetes, it's autonomous. Whatever the workload is, that team can just go their way, and Kubernetes prevents other bottlenecks."

- -

Looking ahead, Jacobs sees Zalando's new infrastructure as a great enabler for other things the company has in the works, from its new logistics software, to a platform feature connecting brands, to products dreamed up by data scientists. "One vision is if you watch the next James Bond movie and see the suit he's wearing, you should be able to automatically order it, and have it delivered to you within an hour," says Jacobs. "It's about connecting the full fashion sphere. This is definitely not possible if you have a bottleneck with everyone running in the same data center and thus very restricted. You need infrastructure to scale each autonomous team's idea."

- -

For other companies considering this technology, Jacobs says he wouldn't necessarily advise doing it exactly the same way Zalando did. "It's okay to do so if you're ready to fail at some things," he says. "You need to set the right expectations. Not everything will work. Rewriting apps and this type of organizational change can be disruptive. The first product we moved was critical. There were a lot of dependencies, and it took longer than expected. Maybe we should have started with something less complicated, less business critical, just to get our toes wet."

- -

But once they got to the other side "it was clear for everyone that there's no big alternative," Jacobs adds. "The Kubernetes API allows us to run applications in a cloud provider-agnostic way, which gives us the freedom to revisit IaaS providers in the coming years. Zalando Technology benefits from migrating to Kubernetes as we are able to leverage our existing knowledge to create an engineering platform offering flexibility and speed to our engineers while significantly reducing the operational overhead. We expect the Kubernetes API to be the global standard for PaaS infrastructure and are excited about the continued journey."

diff --git a/content/ko/case-studies/zalando/zalando_feature_logo.png b/content/ko/case-studies/zalando/zalando_feature_logo.png deleted file mode 100644 index ba6251050d15abb523d009ad79df06cc1fa0779e..0000000000000000000000000000000000000000 GIT binary patch literal 0 HcmV?d00001 literal 7643 zcmd6McQl+`*Y{{4?hv9z%NP+Y%3v@`^xj*dg)jz#QO4*|qnGHti(aF5qIaT4ZxI9$ zJ^Cy6{oMES{qd}KJ?s1H-D_QIowN7;?RL&S*E-jA`{i~OKq>8lFarSOF}C9bb8vA2ghgHK zjA2%AG|&WYj<68{?Kic7fCy6&&?|^Cm$IEC+yWuzhJtIlsc6C6tYA=6kmw7bu#2Dz z(hdnn8v|XC);11;E+U{mjSJq5f3rD3z&};cRw5wr-(G<_%4$GKTNE4!;oxJ3fx%!P zKa>Lu5#Zs5JO^@ffgzk+P);xxI~Q0G%q7Up1^gF*?!2K)%>*^1Wd7yrZYBcyS5nT- z&K%A>9JVNPPB0V-<>cb#HZ6hw&e^9kg8S;G7z82U|xJ41TB1@HhCb zcK^Gh-->t92&$qGcZ*_dEoBRHM8a**@=_w8I|+v=!c>qK1{2`p;x%UH;^X6H=Y^W` zvKw;=aNiY>iyO{uVgfdSL;tq(pYS|T0VxQ#gg6hk3MTY^wV8iV%*LH`~VpLJz@Tvu>PTc z!7(u~?xc6mKZ5^uaQg{Bj0f-ne8IvX24E6nU=d^7b^-G5&JL{Kw*P&s;N8c^y@!MS zn}_vV_KySu6ASMi?tN?=A^_kXCI%KJ7WTb+fBY8TNw9J50dVn%AKjkdbZMGO!HEXiPDcz=4sMfu>~|KWj8`}4QC>uIu1 z8o!odi-OiqEH|tj*B(s@hh@qll^=gB7Lt9fD4G-+RzGPJuRw75{rE$HZv37hX)bF= zk8*VIeQzP<`2}^bRNkW@C6$<#shF2u}53Jqf(+FpM=WEW>2nC+K`tmH^B2% z9?0_URs#5eM{In75|m=tZh&9Qj%WN#a~>2-I-6|#Aww@?gM(6*=8H*$7)+J(oVIn6 zf|ACW`kQ2Hp(Fb{E*Jbv7CCcOR!cCQT(;H;;+&D$kS;5#xv`QiD=>A619?t#?$bpx z-4Ka4SmksFISFsxVt0gehYNT=+#;sS*cw$kTv})%_R@PoBdbfa=LI{fpLiT4>tK28J z=u=L|Q!|U+1>tl9%00nX&r)^Duc20@&(QOo<)kiMg7;h%y9~{?QAImN9(lI_*YxmW zjcu~o6lCon_ev?q>ZEAbAsEq6$RvBk{-sBzS{<$;S}B3IvRt9p{%yUR-unT~(nJe3 zK(c6BBSM_8H5cOYK>gBBFhE)N-1!&Qx8-giuF-x=zNp5j`jbfvLt+|BSLa{P(g!z0 zsgI{id>7}9T&&og3rnAao)*mvdNhLYF-NA`sZzV1{G3ntp)P!xWRZ(8h^8WgPstVz zh%@8)!-#nVrI0U&-z_k$(Y4T@TQ1Z^#A*?;q;-2~|0t|Kqo1m)S&6W1i&oknivxZ#fbdRtn615BuKSNwNx^F6fm&F9ov^@Hk+N0ILCMu@qPky`+#IsZp?V#nHXlo9iwpQ#{< z@R0+AnVG2f-Y|)-J-w7z8bFp!Nou6;<CD9yzPNNYZ(X{^VtiKlG|+l5FU08e zy{SF9>g4`6d-8gjQE47u{6^Dg#o!0`6FW4~;N-cc<*G=7(KRp9&(?JeP=dnai+%lT z1GLEa1~Dx>&T2~HT1i^1jJ$F%Ya!}gOO>?ZWHzl^z#}#*8-MruGkh0+Ti#gU;i_-1 zwg$s_iidjBw7*=>7VW?Qg}w*$5pzW0wcHcB77@GAfwp!ls!;pyE$8Jsm#-9!zRwF8 z%;^`b-CI>ju!7)U5ozr$0txj}msQT+<&1>J5BST5|^}b&ATTDyC?ISlOL> z*d+0#$Oep|lKEm7L_DSJ`IJc!r|EOew)iE*0Lf?uQy;RAhwYGc3O_=d{mf zS9*u0W$ie`FmuTs)C)Z|Ajb4$ z2$r0=Z>iU`D)O`I_lt(eJY^cz@#*ur*qYYYulYul05?E7ONkZA%`@_OdyD7jOzO-J z(IWkr?KGa~;PE5}v*sIt;`bG|fRY=VoYT&+;z_r1?M*XCD-9*flK_3w>RZ6Z6ANwA zjkeP^_iIj{*44|Kt92EJqW!raBPpMq-J7<{pbc*Yq`Y_L+HG-O7 z2UNhr9Ucn(0SclFP>)T8b*rlZ#M}u*uyJPu(B9*f6xY?hTYx=`ug~pREISWb*0QVA`^RhWNX!(|0qG_MN5H4-T;XEbGVN<4@}*P zI~JdOQ|*&n)LlP|h`ZsoV~98e1Pm{6fsZyT7FHJI$oNSH1!@!&eP4$AxxgdyB@%Oa zDA!6s=X5LRMQBS?2;wzTaDMglC(3_a1?4_$= z-%TASYMiC7(0V#)L1ArlP-B%=TkV4^;LUi6|HP`ZpxEq32rARe!q*J^Na@RgxyRe` z6t|a$xL0${jyF1mUIA(T5#-=YlG6i(QQ894nyLU~LxUpWCu2;OrM^bWH0WNrzTPPh3}PQdFt4mFndzD)bfdFlgG`g&D|`e^KCJ3s%MRgWX8i-%;`)TvDixudSUCrpBR&y*y@R$wuzqXdBzan*r0A?b%qP~M zkLPqo)Xizptf^Z@nTTkeT5apz&1laMvz09Czj(Q}f7t`SrovC!KcZA#1AdH*?|hQGH2QMV>E_ z+cVqF6c2N@M>zPihewBAB!`iW+47Lcm4d@V{6Y*!L#>C6nd<;H6B`MPc~N9#-QDMb z2f6FF0Ha?U8)lQ4cDDctH9oqQsT~!~(rR|EhjNL2@iJzQvpF^FwG@D~NxTM>LNu8D zl(<s={mSHht)yHwc+B;%+j$+R%aSbc6OE;)Q&mHb^V$*GRCjGl^yv zkzVygnUQ)+4U*+8v)ArgvxV!czuxH;!D>pdOd&g?i=A>;HMEVO^aFUw*n2e-I;VZ*a=P)GqgdgM$%pW>uRQ$-8xX6Gk zG`B(djQQF*pj;`gjI0n& zh%jXA_Aw>+Vpp0)euCoPCyXFhD~dUj_b!~fnHHhj;oYHZQQ4X7A4FdX?9-$RNll;c z#mDzrPf$POk`dBlDe){Vr6})L&DdOSE`Qxfa6xV>*JS(bg>B*H;Ct7rxYbr+((dR6 zD_{0n(fZWboMq+uLh{^J_aTh~hc+YbFT;oKL&J+3o3TSzvcI-sdG@5Kd2RvK?&I|4 zY?_Rbr33uzex~N@LSq~$GZB4B+FBvhzlu7TA!hh_GQKw+qhku)Y`w@^!1swdTmjA8 zU%0&?UbNM*uloFafcp$s;?7YnRwuCR>+PN%e{Crdo zPRVdkc?LwmQ}AQoPAYas4CS}*0fsHH!4LQQT@TdxLdW&;-!JV5TubNFToHbpt_jiV zi2II&7wxWo25f7zc1pdrX zJ09pjEw5(>{Hn@#4DdbE(yw`zocU>o7O9@EOI$Ku%x=Kr7p$Ne^idOqUum=L%o=l~ zi(p)0`F!nL^z9Lb0#GeHMsRi7`4+%0vw90SFKO_(DyAo?fom<@wR>+RPt@<$xmM29EQz%2#!3wUadwve>e zr#_|2wYL;$--hz|Lu_n)6kvFWga9Ju`+oR}*!M?lyfH;`Sfr#pKM^+Q9rV-%AXw1B z#l6*@bFn0FI%x5(d{fo*R^)e9u9`BPN@szUTzVyus4-dX@w_Jzpx~hwJ~8ceI!8Cl z_q;t?65RGCx&p}_`NUYB4};VDb99q4D=h{5KX&a|6;Jf^l1S#8S&rg=C{bv<#KQ@Pk6f-Cfp6h6+jsa73s`*Gsq0BE5HM+;7V zIztt*$$Vc^{X5VmZ8chV^`g!9%7t1ZDf6dofR-Y)G;Ei*jQLYF*ew-`2dYlB|tiRxP17sV!B zty!(%UN0HyvR6j->SPp6hV)KuEnC0HKW!~lf zNpLKRY%Mz(8dF3s?|xL;hC)+gdvz{H3Z*CufO9Asr3*G3@}Uqz*)qPJ~8mO-_8ekHseoS5>QML{Cgk3~*` z6kmcAkjtV#i?0zYh@qfH=#BAnq#!7y>t#hQ3@tD+T~o?eJv3*oo3zZmDfHgAb$U`? z^A+S{OF_e|;L?FO(kfhOM{b&;ewvZ3;EN6@rWT*|luVuo6{XPl3_K3N!4nN)A<6mP zH=T3Zf!Y-Je|Pczu*!TQu$GA?E=&1!c`X5f6+Rdrs>gCapUh*Cv^vFQBb^nksiPt_ zk`<!<6w182}gDw8Fpjj2;Xb(=>lOSeu1Ss1k2rbrA z!sc396F!hx=;C@5RkU;9Hm~FAC!bzZRwY4Izod0ZIv%Bl#3PvHdb1?MG5HN#OrFa^ zWF4EC;?D&iCqdhFHtYKn+Lw^Yh@Bh#-+SNTY@G%bdm};#T zKt^#yLNcw-o<3&xb~-zItVBair^ifsxmeB^w2lq7&A9t~np`whB&4qDq0Ln~dMAUs6j~uJgtr3wn9Di@_FS$@xHjwQjM# z#CaxCyDkJ_UEY;U-}el5_x%3srm2zznMvP@-pWbrOj^G%p?P-CT+9_igP!s2s&CIR z8?y_q>L<0;wCR=T=U{Xs9*Fkrt;I~jfGbUzA>}WZ+UQb7v94TEL{ ztt9U%`bB3Oa|MUz8eBxq`N}B11n<_xaN4$|eKe7xtPV@&Jqp;amK|#XDudOf@!=7A z-;Vg!4%9j`Y*jg3nmp=mxRB?z(PZ^H}d+L#H-BI68$pS>()Wsg zBwN*(ye5}KrpS^YIgdd4?AG|c*VW6!QAD9w~CXCo+_g$hE4G4WXhAOgVJ R?Uet&{~zE#M<{wb`#*xsh{XT^ From 64558544ec824e7e2dc5cb0077bbbe7d6c47e7d0 Mon Sep 17 00:00:00 2001 From: seokho-son Date: Sun, 30 May 2021 18:49:25 +0900 Subject: [PATCH 06/10] Update outdated files in dev-1.21-ko.3(M28-M36) --- ...port-forward-access-application-cluster.md | 5 +-- .../web-ui-dashboard.md | 7 +++- .../kubeadm/kubeadm-certs.md | 2 +- .../job/automated-tasks-with-cron-jobs.md | 8 ++-- .../kustomization.md | 37 +++++++++++++++++++ .../run-application/delete-stateful-set.md | 12 ++++-- .../examples/application/job/redis/worker.py | 4 +- .../ko/examples/controllers/daemonset.yaml | 1 + content/ko/includes/task-tutorial-prereqs.md | 4 +- 9 files changed, 64 insertions(+), 16 deletions(-) diff --git a/content/ko/docs/tasks/access-application-cluster/port-forward-access-application-cluster.md b/content/ko/docs/tasks/access-application-cluster/port-forward-access-application-cluster.md index cc5f872cc5..333001af81 100644 --- a/content/ko/docs/tasks/access-application-cluster/port-forward-access-application-cluster.md +++ b/content/ko/docs/tasks/access-application-cluster/port-forward-access-application-cluster.md @@ -111,7 +111,6 @@ min-kubernetes-server-version: v1.10 ```shell # mongo-75f59d57f4-4nd6q 를 당신의 파드 이름으로 대체한다. kubectl get pod mongo-75f59d57f4-4nd6q --template='{{(index (index .spec.containers 0).ports 0).containerPort}}{{"\n"}}' - ``` 출력은 파드 내 MongoDB 포트 번호를 보여준다. @@ -124,7 +123,7 @@ min-kubernetes-server-version: v1.10 ## 파드의 포트를 로컬 포트로 포워딩하기 -1. `kubectl port-forward` 명령어는 파드 이름과 같이 리소스 이름을 사용하여 일치하는 파드를 선택해 포트 포워딩하는 것을 허용한다. +1. `kubectl port-forward` 명령어는 파드 이름과 같이 리소스 이름을 사용하여 일치하는 파드를 선택해 포트 포워딩하는 것을 허용한다. ```shell @@ -177,7 +176,7 @@ min-kubernetes-server-version: v1.10 3. MongoDB 커맨드라인 프롬프트에 `ping` 명령을 입력한다. - ```shell + ``` db.runCommand( { ping: 1 } ) ``` diff --git a/content/ko/docs/tasks/access-application-cluster/web-ui-dashboard.md b/content/ko/docs/tasks/access-application-cluster/web-ui-dashboard.md index d0cfee3009..7cb694386d 100644 --- a/content/ko/docs/tasks/access-application-cluster/web-ui-dashboard.md +++ b/content/ko/docs/tasks/access-application-cluster/web-ui-dashboard.md @@ -1,4 +1,8 @@ --- + + + + title: 웹 UI (대시보드) content_type: concept weight: 10 @@ -30,12 +34,11 @@ card: 대시보드 UI는 기본으로 배포되지 않는다. 배포하려면 다음 커맨드를 동작한다. ``` -kubectl apply -f https://raw.githubusercontent.com/kubernetes/dashboard/v2.0.0/aio/deploy/recommended.yaml +kubectl apply -f https://raw.githubusercontent.com/kubernetes/dashboard/v2.2.0/aio/deploy/recommended.yaml ``` ## 대시보드 UI 접근 - 클러스터 데이터를 보호하기 위해, 대시보드는 기본적으로 최소한의 RBAC 설정을 제공한다. 현재, 대시보드는 Bearer 토큰으로 로그인 하는 방법을 제공한다. 본 시연을 위한 토큰을 생성하기 위해서는, diff --git a/content/ko/docs/tasks/administer-cluster/kubeadm/kubeadm-certs.md b/content/ko/docs/tasks/administer-cluster/kubeadm/kubeadm-certs.md index 78e2fa767b..02b2b1330f 100644 --- a/content/ko/docs/tasks/administer-cluster/kubeadm/kubeadm-certs.md +++ b/content/ko/docs/tasks/administer-cluster/kubeadm/kubeadm-certs.md @@ -229,7 +229,7 @@ serverTLSBootstrap: true 만약 이미 클러스터를 생성했다면 다음을 따라 이를 조정해야 한다. - `kube-system` 네임스페이스에서 `kubelet-config-{{< skew latestVersion >}}` 컨피그맵을 찾아서 수정한다. -해당 컨피그맵에는 `config` 키가 +해당 컨피그맵에는 `kubelet` 키가 [KubeletConfiguration](/docs/reference/config-api/kubelet-config.v1beta1/#kubelet-config-k8s-io-v1beta1-KubeletConfiguration) 문서를 값으로 가진다. `serverTLSBootstrap: true` 가 되도록 KubeletConfiguration 문서를 수정한다. - 각 노드에서, `serverTLSBootstrap: true` 필드를 `/var/lib/kubelet/config.yaml` 에 추가한다. diff --git a/content/ko/docs/tasks/job/automated-tasks-with-cron-jobs.md b/content/ko/docs/tasks/job/automated-tasks-with-cron-jobs.md index 4addcdcfaf..26353959ef 100644 --- a/content/ko/docs/tasks/job/automated-tasks-with-cron-jobs.md +++ b/content/ko/docs/tasks/job/automated-tasks-with-cron-jobs.md @@ -1,6 +1,8 @@ --- title: 크론잡(CronJob)으로 자동화된 작업 실행 min-kubernetes-server-version: v1.21 + + content_type: task weight: 10 --- @@ -144,7 +146,7 @@ kubectl delete cronjob hello `.spec.schedule` 은 `.spec` 의 필수 필드이다. 이는 해당 잡이 생성되고 실행되는 스케줄 시간으로 `0 * * * *` 또는 `@hourly` 와 같이 [크론](https://ko.wikipedia.org/wiki/Cron) 형식의 문자열을 받아들인다. -이 형식은 확장된 `vixie cron` 스텝(step) 값도 포함한다. 이 내용은 +이 형식은 확장된 "Vixie cron" 스텝(step) 값도 포함한다. 이 내용은 [FreeBSD 매뉴얼](https://www.freebsd.org/cgi/man.cgi?crontab%285%29)에 설명되어 있다. > 스텝 값은 범위(range)와 함께 사용할 수 있다. 범위 뒤에 `/` 를 @@ -172,8 +174,8 @@ kubectl delete cronjob hello 이러한 방식으로 기한을 맞추지 못한 잡은 실패한 작업으로 간주된다. 이 필드를 지정하지 않으면, 잡에 기한이 없다. -`.spec.startingDeadlineSeconds` 필드가 (null이 아닌 값으로) 설정되어 있다면, -크론잡 컨트롤러는 잡 생성 완료 예상 시각과 현재 시각의 차이를 측정하고, +`.spec.startingDeadlineSeconds` 필드가 (null이 아닌 값으로) 설정되어 있다면, +크론잡 컨트롤러는 잡 생성 완료 예상 시각과 현재 시각의 차이를 측정하고, 시각 차이가 설정한 값보다 커지면 잡 생성 동작을 스킵한다. 예를 들어, `200` 으로 설정되었다면, 잡 생성 완료 예상 시각으로부터 200초까지는 잡이 생성될 수 있다. diff --git a/content/ko/docs/tasks/manage-kubernetes-objects/kustomization.md b/content/ko/docs/tasks/manage-kubernetes-objects/kustomization.md index feff78ce25..ab442ebafd 100644 --- a/content/ko/docs/tasks/manage-kubernetes-objects/kustomization.md +++ b/content/ko/docs/tasks/manage-kubernetes-objects/kustomization.md @@ -86,6 +86,43 @@ metadata: name: example-configmap-1-8mbdf7882g ``` +env 파일에서 컨피그맵을 생성하려면, `configMapGenerator`의 `envs` 리스트에 항목을 추가한다. 다음은 `.env` 파일의 데이터 항목으로 컨피그맵을 생성하는 예시를 보여준다. + +```shell +# .env 파일 생성 +cat <.env +FOO=Bar +EOF + +cat <./kustomization.yaml +configMapGenerator: +- name: example-configmap-1 + envs: + - .env +EOF +``` + +생성된 컨피그맵은 다음 명령어로 검사할 수 있다. + +```shell +kubectl kustomize ./ +``` + +생성된 컨피그맵은 다음과 같다. + +```yaml +apiVersion: v1 +data: + FOO=Bar +kind: ConfigMap +metadata: + name: example-configmap-1-8mbdf7882g +``` + +{{< note >}} +`.env` 파일의 각 변수는 생성한 컨피그맵에서 분리된 키가 된다. `.properties` 라는 이름의 파일을 내장하는 이전 예시(그리고 모든 항목들)는 단일 키를 위한 값이므로 이 예시와는 다르다. +{{< /note >}} + 컨피그맵은 문자로된 키-값 쌍들로도 생성할 수 있다. 문자로된 키-값 쌍에서 컨피그맵을 생성하려면, configMapGenerator 내의 `literals` 리스트에 항목을 추가한다. 다음은 키-값 쌍을 데이터 항목으로 받는 컨피그맵을 생성하는 예제이다. ```shell diff --git a/content/ko/docs/tasks/run-application/delete-stateful-set.md b/content/ko/docs/tasks/run-application/delete-stateful-set.md index 07b3396440..7c2b1ed783 100644 --- a/content/ko/docs/tasks/run-application/delete-stateful-set.md +++ b/content/ko/docs/tasks/run-application/delete-stateful-set.md @@ -1,4 +1,10 @@ --- + + + + + + title: 스테이트풀셋(StatefulSet) 삭제하기 content_type: task weight: 60 @@ -37,14 +43,14 @@ kubectl delete statefulsets kubectl delete service ``` -kubectl을 통해 스테이트풀셋을 삭제하면, 스테이트풀셋의 크기가 0으로 설정되고 이로 인해 스테이트풀셋에 포함된 모든 파드가 삭제된다. 파드가 아닌 스테이트풀셋만 삭제하려면, `--cascade=false` 옵션을 사용한다. +kubectl을 통해 스테이트풀셋을 삭제하면, 스테이트풀셋의 크기가 0으로 설정되고 이로 인해 스테이트풀셋에 포함된 모든 파드가 삭제된다. 파드가 아닌 스테이트풀셋만 삭제하려면, `--cascade=orphan` 옵션을 사용한다. 예시는 다음과 같다. ```shell -kubectl delete -f --cascade=false +kubectl delete -f --cascade=orphan ``` -`kubectl delete` 에 `--cascade=false` 를 사용하면 스테이트풀셋 오브젝트가 삭제된 후에도 스테이트풀셋에 의해 관리된 파드는 남게 된다. 만약 파드가 `app=myapp` 레이블을 갖고 있다면, 다음과 같이 파드를 삭제할 수 있다. +`kubectl delete` 에 `--cascade=orphan` 를 사용하면 스테이트풀셋 오브젝트가 삭제된 후에도 스테이트풀셋에 의해 관리된 파드는 남게 된다. 만약 파드가 `app=myapp` 레이블을 갖고 있다면, 다음과 같이 파드를 삭제할 수 있다. ```shell kubectl delete pods -l app=myapp diff --git a/content/ko/examples/application/job/redis/worker.py b/content/ko/examples/application/job/redis/worker.py index b8abbee917..0e24f71f95 100644 --- a/content/ko/examples/application/job/redis/worker.py +++ b/content/ko/examples/application/job/redis/worker.py @@ -8,11 +8,11 @@ host="redis" # import os # host = os.getenv("REDIS_SERVICE_HOST") -q = rediswq.RedisWQ(name="job2", host="redis") +q = rediswq.RedisWQ(name="job2", host=host) print("Worker with sessionID: " + q.sessionID()) print("Initial queue state: empty=" + str(q.empty())) while not q.empty(): - item = q.lease(lease_secs=10, block=True, timeout=2) + item = q.lease(lease_secs=10, block=True, timeout=2) if item is not None: itemstr = item.decode("utf-8") print("Working on " + itemstr) diff --git a/content/ko/examples/controllers/daemonset.yaml b/content/ko/examples/controllers/daemonset.yaml index f291b750c1..685a137244 100644 --- a/content/ko/examples/controllers/daemonset.yaml +++ b/content/ko/examples/controllers/daemonset.yaml @@ -18,6 +18,7 @@ spec: # this toleration is to have the daemonset runnable on master nodes # remove it if your masters can't run pods - key: node-role.kubernetes.io/master + operator: Exists effect: NoSchedule containers: - name: fluentd-elasticsearch diff --git a/content/ko/includes/task-tutorial-prereqs.md b/content/ko/includes/task-tutorial-prereqs.md index 90aa725d6c..65651286bd 100644 --- a/content/ko/includes/task-tutorial-prereqs.md +++ b/content/ko/includes/task-tutorial-prereqs.md @@ -1,6 +1,6 @@ 쿠버네티스 클러스터가 필요하고, kubectl 커맨드-라인 툴이 클러스터와 -통신할 수 있도록 설정되어 있어야 한다. -만약, 아직 클러스터를 가지고 있지 않다면, +통신할 수 있도록 설정되어 있어야 한다. 이 튜토리얼은 컨트롤 플레인 호스트가 아닌 노드가 적어도 2개 포함된 클러스터에서 실행하는 것을 추천한다. 만약, 아직 클러스터를 가지고 +있지 않다면, [minikube](/ko/docs/tasks/tools/#minikube)를 사용해서 생성하거나 다음의 쿠버네티스 플레이그라운드 중 하나를 사용할 수 있다. From 1ada7f1a26140c16ad661a50713bf897093f47d4 Mon Sep 17 00:00:00 2001 From: jmyung Date: Mon, 24 May 2021 23:02:10 +0900 Subject: [PATCH 07/10] Translate docs/tasks/configure-pod-container/configure-runasusername in Korean --- .../configure-runasusername.md | 126 ++++++++++++++++++ .../ko/examples/windows/configmap-pod.yaml | 31 +++++ content/ko/examples/windows/daemonset.yaml | 21 +++ .../ko/examples/windows/deploy-hyperv.yaml | 22 +++ .../ko/examples/windows/deploy-resource.yaml | 24 ++++ content/ko/examples/windows/emptydir-pod.yaml | 20 +++ .../examples/windows/hostpath-volume-pod.yaml | 18 +++ .../windows/run-as-username-container.yaml | 17 +++ .../examples/windows/run-as-username-pod.yaml | 14 ++ content/ko/examples/windows/secret-pod.yaml | 32 +++++ content/ko/examples/windows/simple-pod.yaml | 14 ++ 11 files changed, 339 insertions(+) create mode 100644 content/ko/docs/tasks/configure-pod-container/configure-runasusername.md create mode 100644 content/ko/examples/windows/configmap-pod.yaml create mode 100644 content/ko/examples/windows/daemonset.yaml create mode 100644 content/ko/examples/windows/deploy-hyperv.yaml create mode 100644 content/ko/examples/windows/deploy-resource.yaml create mode 100644 content/ko/examples/windows/emptydir-pod.yaml create mode 100644 content/ko/examples/windows/hostpath-volume-pod.yaml create mode 100644 content/ko/examples/windows/run-as-username-container.yaml create mode 100644 content/ko/examples/windows/run-as-username-pod.yaml create mode 100644 content/ko/examples/windows/secret-pod.yaml create mode 100644 content/ko/examples/windows/simple-pod.yaml diff --git a/content/ko/docs/tasks/configure-pod-container/configure-runasusername.md b/content/ko/docs/tasks/configure-pod-container/configure-runasusername.md new file mode 100644 index 0000000000..8bae2f4286 --- /dev/null +++ b/content/ko/docs/tasks/configure-pod-container/configure-runasusername.md @@ -0,0 +1,126 @@ +--- +title: 윈도우 파드 및 컨테이너에서 RunAsUserName 구성 +content_type: task +weight: 20 +--- + + + +{{< feature-state for_k8s_version="v1.18" state="stable" >}} + +이 페이지에서는 윈도우 노드에서 실행될 파드 및 컨테이너에 `runAsUserName` 설정을 사용하는 방법을 소개한다. 이는 리눅스 관련 `runAsUser` 설정과 거의 동일하여, 컨테이너의 기본값과 다른 username으로 애플리케이션을 실행할 수 있다. + + + +## {{% heading "prerequisites" %}} + + +쿠버네티스 클러스터가 있어야 하며 클러스터와 통신하도록 kubectl 명령줄 도구를 구성해야 한다. 클러스터에는 윈도우 워커 노드가 있어야 하고, 해당 노드에서 윈도우 워크로드를 실행하는 컨테이너의 파드가 스케쥴 된다. + + + + + +## 파드의 username 설정 + +파드의 컨테이너 프로세스를 실행할 username을 지정하려면 파드 명세에 `securityContext` 필드 ([PodSecurityContext](/docs/reference/generated/kubernetes-api/{{< param "version" >}}/#podsecuritycontext-v1-core)) 를 포함시키고, 그 안에 `runAsUserName` 필드를 포함하는 `windowsOptions` ([WindowsSecurityContextOptions](/docs/reference/generated/kubernetes-api/{{< param "version" >}}/#windowssecuritycontextoptions-v1-core)) 필드를 추가한다. + +파드에 지정하는 윈도우 보안 컨텍스트 옵션은 파드의 모든 컨테이너 및 초기화 컨테이너에 적용된다. + +다음은 `runAsUserName` 필드가 설정된 윈도우 파드의 구성 파일이다. + +{{< codenew file="windows/run-as-username-pod.yaml" >}} + +파드를 생성한다. + +```shell +kubectl apply -f https://k8s.io/examples/windows/run-as-username-pod.yaml +``` + +파드의 컨테이너가 실행 중인지 확인한다. + +```shell +kubectl get pod run-as-username-pod-demo +``` + +실행 중인 컨테이너의 셸에 접근한다. + +```shell +kubectl exec -it run-as-username-pod-demo -- powershell +``` + +셸이 올바른 username인 사용자로 실행 중인지 확인한다. + +```powershell +echo $env:USERNAME +``` + +결과는 다음과 같다. + +```shell +ContainerUser +``` + +## 컨테이너의 username 설정 + +컨테이너의 프로세스를 실행할 username을 지정하려면, 컨테이너 매니페스트에 `securityContext` 필드 ([SecurityContext](/docs/reference/generated/kubernetes-api/{{< param "version" >}}/#securitycontext-v1-core)) 를 포함시키고 그 안에 `runAsUserName` 필드를 포함하는 `windowsOptions` ([WindowsSecurityContextOptions](/docs/reference/generated/kubernetes-api/{{< param "version" >}}/#windowssecuritycontextoptions-v1-core)) 필드를 추가한다. + +컨테이너에 지정하는 윈도우 보안 컨텍스트 옵션은 해당 개별 컨테이너에만 적용되며 파드 수준에서 지정한 설정을 재정의한다. + +다음은 한 개의 컨테이너에 `runAsUserName` 필드가 파드 수준 및 컨테이너 수준에서 설정되는 파드의 구성 파일이다. + +{{< codenew file="windows/run-as-username-container.yaml" >}} + +파드를 생성한다. + +```shell +kubectl apply -f https://k8s.io/examples/windows/run-as-username-container.yaml +``` + +파드의 컨테이너가 실행 중인지 확인한다. + +```shell +kubectl get pod run-as-username-container-demo +``` + +실행 중인 컨테이너의 셸에 접근한다. + +```shell +kubectl exec -it run-as-username-container-demo -- powershell +``` + +셸이 사용자에게 올바른 username(컨테이너 수준에서 설정된 사용자)을 실행 중인지 확인한다. + +```powershell +echo $env:USERNAME +``` + +결과는 다음과 같다. + +```shell +ContainerAdministrator +``` + +## 윈도우 username 제약사항 + +이 기능을 사용하려면 `runAsUserName` 필드에 설정된 값이 유효한 username이어야 한다. 형식은 `DOMAIN\USER` 여야하고, 여기서 `DOMAIN\`은 선택 사항이다. 윈도우 username은 대소문자를 구분하지 않는다. 또한 `DOMAIN` 및 `USER` 와 관련된 몇 가지 제약사항이 있다. + +- `runAsUserName` 필드는 비워 둘 수 없으며 제어 문자를 포함할 수 없다. (ASCII 값: `0x00-0x1F`, `0x7F`) +- `DOMAIN`은 NetBios 이름 또는 DNS 이름이어야 하며 각각 고유한 제한이 있다. + - NetBios 이름: 최대 15 자, `.`(마침표)으로 시작할 수 없으며 다음 문자를 포함할 수 없다. `\ / : * ? " < > |` + - DNS 이름: 최대 255 자로 영숫자, 마침표(`.`), 대시(`-`)로만 구성되며, 마침표 또는 대시로 시작하거나 끝날 수 없다. +- `USER`는 최대 20자이며, *오직* 마침표나 공백들로는 구성할 수 없고, 다음 문자는 포함할 수 없다. `" / \ [ ] : ; | = , + * ? < > @`. + +`runAsUserName` 필드에 허용되는 값의 예 : `ContainerAdministrator`,`ContainerUser`, `NT AUTHORITY\NETWORK SERVICE`, `NT AUTHORITY\LOCAL SERVICE`. + +이러한 제약사항에 대한 자세한 내용은 [여기](https://support.microsoft.com/en-us/help/909264/naming-conventions-in-active-directory-for-computers-domains-sites-and) 와 [여기](https://docs.microsoft.com/en-us/powershell/module/microsoft.powershell.localaccounts/new-localuser?view=powershell-5.1)를 확인한다. + + + +## {{% heading "whatsnext" %}} + + +* [쿠버네티스에서 윈도우 컨테이너 스케줄링을 위한 가이드](/ko/docs/setup/production-environment/windows/user-guide-windows-containers/) +* [그룹 매니지드 서비스 어카운트를 이용하여 워크로드 신원 관리하기](/ko/docs/setup/production-environment/windows/user-guide-windows-containers/#그룹-매니지드-서비스-어카운트를-이용하여-워크로드-신원-관리하기) +* [윈도우 파드와 컨테이너의 GMSA 구성](/docs/tasks/configure-pod-container/configure-gmsa/) + diff --git a/content/ko/examples/windows/configmap-pod.yaml b/content/ko/examples/windows/configmap-pod.yaml new file mode 100644 index 0000000000..661cb73dee --- /dev/null +++ b/content/ko/examples/windows/configmap-pod.yaml @@ -0,0 +1,31 @@ +kind: ConfigMap +apiVersion: v1 +metadata: + name: example-config +data: + example.property.1: hello + example.property.2: world + +--- + +apiVersion: v1 +kind: Pod +metadata: + name: configmap-pod +spec: + containers: + - name: configmap-redis + image: redis:3.0-nanoserver + env: + - name: EXAMPLE_PROPERTY_1 + valueFrom: + configMapKeyRef: + name: example-config + key: example.property.1 + - name: EXAMPLE_PROPERTY_2 + valueFrom: + configMapKeyRef: + name: example-config + key: example.property.2 + nodeSelector: + kubernetes.io/os: windows \ No newline at end of file diff --git a/content/ko/examples/windows/daemonset.yaml b/content/ko/examples/windows/daemonset.yaml new file mode 100644 index 0000000000..7483708fc7 --- /dev/null +++ b/content/ko/examples/windows/daemonset.yaml @@ -0,0 +1,21 @@ +apiVersion: apps/v1 +kind: DaemonSet +metadata: + name: my-daemonset + labels: + app: foo +spec: + selector: + matchLabels: + app: foo + template: + metadata: + labels: + app: foo + spec: + containers: + - name: foo + image: microsoft/windowsservercore:1709 + nodeSelector: + kubernetes.io/os: windows + diff --git a/content/ko/examples/windows/deploy-hyperv.yaml b/content/ko/examples/windows/deploy-hyperv.yaml new file mode 100644 index 0000000000..c8b71ce8cb --- /dev/null +++ b/content/ko/examples/windows/deploy-hyperv.yaml @@ -0,0 +1,22 @@ +apiVersion: apps/v1 +kind: Deployment +metadata: + name: iis +spec: + selector: + matchLabels: + app: iis + replicas: 3 + template: + metadata: + labels: + app: iis + annotations: + experimental.windows.kubernetes.io/isolation-type: hyperv + spec: + containers: + - name: iis + image: microsoft/iis + ports: + - containerPort: 80 + diff --git a/content/ko/examples/windows/deploy-resource.yaml b/content/ko/examples/windows/deploy-resource.yaml new file mode 100644 index 0000000000..81207a3804 --- /dev/null +++ b/content/ko/examples/windows/deploy-resource.yaml @@ -0,0 +1,24 @@ +apiVersion: apps/v1 +kind: Deployment +metadata: + name: iis +spec: + replicas: 3 + selector: + matchLabels: + app: iis + template: + metadata: + labels: + app: iis + spec: + containers: + - name: iis + image: microsoft/iis + resources: + limits: + memory: "128Mi" + cpu: 2 + ports: + - containerPort: 80 + diff --git a/content/ko/examples/windows/emptydir-pod.yaml b/content/ko/examples/windows/emptydir-pod.yaml new file mode 100644 index 0000000000..08d8091391 --- /dev/null +++ b/content/ko/examples/windows/emptydir-pod.yaml @@ -0,0 +1,20 @@ +apiVersion: v1 +kind: Pod +metadata: + name: my-empty-dir-pod +spec: + containers: + - image: microsoft/windowsservercore:1709 + name: my-empty-dir-pod + volumeMounts: + - mountPath: /cache + name: cache-volume + - mountPath: C:/scratch + name: scratch-volume + volumes: + - name: cache-volume + emptyDir: {} + - name: scratch-volume + emptyDir: {} + nodeSelector: + kubernetes.io/os: windows diff --git a/content/ko/examples/windows/hostpath-volume-pod.yaml b/content/ko/examples/windows/hostpath-volume-pod.yaml new file mode 100644 index 0000000000..d95e345b6c --- /dev/null +++ b/content/ko/examples/windows/hostpath-volume-pod.yaml @@ -0,0 +1,18 @@ +apiVersion: v1 +kind: Pod +metadata: + name: hostpath-volume-pod +spec: + containers: + - name: my-hostpath-volume-pod + image: microsoft/windowsservercore:1709 + volumeMounts: + - name: foo + mountPath: "C:\\etc\\foo" + readOnly: true + nodeSelector: + kubernetes.io/os: windows + volumes: + - name: foo + hostPath: + path: "C:\\etc\\foo" diff --git a/content/ko/examples/windows/run-as-username-container.yaml b/content/ko/examples/windows/run-as-username-container.yaml new file mode 100644 index 0000000000..77b7b2d188 --- /dev/null +++ b/content/ko/examples/windows/run-as-username-container.yaml @@ -0,0 +1,17 @@ +apiVersion: v1 +kind: Pod +metadata: + name: run-as-username-container-demo +spec: + securityContext: + windowsOptions: + runAsUserName: "ContainerUser" + containers: + - name: run-as-username-demo + image: mcr.microsoft.com/windows/servercore:ltsc2019 + command: ["ping", "-t", "localhost"] + securityContext: + windowsOptions: + runAsUserName: "ContainerAdministrator" + nodeSelector: + kubernetes.io/os: windows diff --git a/content/ko/examples/windows/run-as-username-pod.yaml b/content/ko/examples/windows/run-as-username-pod.yaml new file mode 100644 index 0000000000..281bbda597 --- /dev/null +++ b/content/ko/examples/windows/run-as-username-pod.yaml @@ -0,0 +1,14 @@ +apiVersion: v1 +kind: Pod +metadata: + name: run-as-username-pod-demo +spec: + securityContext: + windowsOptions: + runAsUserName: "ContainerUser" + containers: + - name: run-as-username-demo + image: mcr.microsoft.com/windows/servercore:ltsc2019 + command: ["ping", "-t", "localhost"] + nodeSelector: + kubernetes.io/os: windows diff --git a/content/ko/examples/windows/secret-pod.yaml b/content/ko/examples/windows/secret-pod.yaml new file mode 100644 index 0000000000..69ee9b1f1e --- /dev/null +++ b/content/ko/examples/windows/secret-pod.yaml @@ -0,0 +1,32 @@ +apiVersion: v1 +kind: Secret +metadata: + name: mysecret +type: Opaque +data: + username: YWRtaW4= + password: MWYyZDFlMmU2N2Rm + +--- + +apiVersion: v1 +kind: Pod +metadata: + name: my-secret-pod +spec: + containers: + - name: my-secret-pod + image: microsoft/windowsservercore:1709 + env: + - name: USERNAME + valueFrom: + secretKeyRef: + name: mysecret + key: username + - name: PASSWORD + valueFrom: + secretKeyRef: + name: mysecret + key: password + nodeSelector: + kubernetes.io/os: windows diff --git a/content/ko/examples/windows/simple-pod.yaml b/content/ko/examples/windows/simple-pod.yaml new file mode 100644 index 0000000000..0b1f0ed5c5 --- /dev/null +++ b/content/ko/examples/windows/simple-pod.yaml @@ -0,0 +1,14 @@ +apiVersion: v1 +kind: Pod +metadata: + name: iis + labels: + name: iis +spec: + containers: + - name: iis + image: microsoft/iis:windowsservercore-1709 + ports: + - containerPort: 80 + nodeSelector: + "kubernetes.io/os": windows From f0be2cf9a5a168ee2a7beb66d456d4c6585a0e92 Mon Sep 17 00:00:00 2001 From: Jihoon Seo Date: Mon, 31 May 2021 18:11:27 +0900 Subject: [PATCH 08/10] [ko] Update outdated files in dev-1.21-ko.3 (p6) --- .../setup/production-environment/_index.md | 291 +++++++++++++++++- .../tools/kubeadm/control-plane-flags.md | 10 +- 2 files changed, 297 insertions(+), 4 deletions(-) diff --git a/content/ko/docs/setup/production-environment/_index.md b/content/ko/docs/setup/production-environment/_index.md index 5296cfcaf2..3471214564 100644 --- a/content/ko/docs/setup/production-environment/_index.md +++ b/content/ko/docs/setup/production-environment/_index.md @@ -1,4 +1,293 @@ --- -title: 운영 환경 +title: "프로덕션 환경" +description: 프로덕션 수준의 쿠버네티스 클러스터 생성 weight: 30 +no_list: true --- + + +프로덕션 수준의 쿠버네티스 클러스터에는 계획과 준비가 필요하다. +쿠버네티스 클러스터에 중요한 워크로드를 실행하려면 클러스터를 탄력적이도록 구성해야 한다. +이 페이지에서는 프로덕션용 클러스터를 설정하거나 기존 클러스터를 프로덕션용으로 업그레이드하기 위해 +수행할 수 있는 단계를 설명한다. +이미 프로덕션 구성 내용에 익숙하여 단지 링크를 찾고 있다면, +[다음 내용](#다음-내용)을 참고한다. + + + +## 프로덕션 고려 사항 + +일반적으로 프로덕션 쿠버네티스 클러스터 환경에는 +개인 학습용, 개발용 또는 테스트 환경용 클러스터보다 더 많은 요구 사항이 있다. +프로덕션 환경에는 많은 사용자의 보안 액세스, 일관된 가용성 및 +변화하는 요구를 충족하기 위한 리소스가 필요할 수 있다. + +프로덕션 쿠버네티스 환경이 상주할 위치(온 프레미스 또는 클라우드)와 +직접 처리하거나 다른 사람에게 맡길 관리의 양을 결정할 때, +쿠버네티스 클러스터에 대한 요구 사항이 +다음 이슈에 의해 어떻게 영향을 받는지 고려해야 한다. + +- *가용성*: 단일 머신 쿠버네티스 [학습 환경](/ko/docs/setup/#학습-환경)은 SPOF(Single Point of Failure, 단일 장애 지점) 이슈를 갖고 있다. +고가용성 클러스터를 만드는 것에는 다음과 같은 고려 사항이 있다. + - 컨트롤 플레인과 워크 노드를 분리 + - 컨트롤 플레인 구성요소를 여러 노드에 복제 + - 클러스터의 {{< glossary_tooltip term_id="kube-apiserver" text="API 서버" >}}로 가는 트래픽을 로드밸런싱 + - 워커 노드를 충분히 운영하거나, 워크로드 변경에 따라 빠르게 제공할 수 있도록 보장 + +- *스케일링*: 프로덕션 쿠버네티스 환경에 들어오는 요청의 양의 +일정할 것으로 예상된다면, 필요한 만큼의 용량(capacity)을 증설하고 +마무리할 수도 있다. 하지만, 요청의 양이 시간에 따라 점점 증가하거나 +계절, 이벤트 등에 의해 극적으로 변동할 것으로 예상된다면, +컨트롤 플레인과 워커 노드로의 요청 증가로 인한 압박을 해소하기 위해 스케일 업 하거나 +잉여 자원을 줄이기 위해 스케일 다운 하는 것에 대해 고려해야 한다. + +- *보안 및 접근 관리*: 학습을 위한 쿠버네티스 클러스터에는 +완전한 관리 권한을 가질 수 있다. 하지만 중요한 워크로드를 실행하며 +두 명 이상의 사용자가 있는 공유 클러스터에는 누가, 그리고 무엇이 클러스터 자원에 +접근할 수 있는지에 대해서 보다 정교한 접근 방식이 필요하다. +역할 기반 접근 제어([RBAC](/docs/reference/access-authn-authz/rbac/)) 및 +기타 보안 메커니즘을 사용하여, 사용자와 워크로드가 필요한 자원에 +액세스할 수 있게 하면서도 워크로드와 클러스터를 안전하게 유지할 수 있다. +[정책](/ko/docs/concepts/policy/)과 +[컨테이너 리소스](/ko/docs/concepts/configuration/manage-resources-containers/)를 +관리하여, 사용자 및 워크로드가 접근할 수 있는 자원에 대한 제한을 설정할 수 있다. + +쿠버네티스 프로덕션 환경을 직접 구축하기 전에, 이 작업의 일부 또는 전체를 +[턴키 클라우드 솔루션](/docs/setup/production-environment/turnkey-solutions/) +제공 업체 또는 기타 [쿠버네티스 파트너](/ko/partners/)에게 +넘기는 것을 고려할 수 있다. +다음과 같은 옵션이 있다. + +- *서버리스*: 클러스터를 전혀 관리하지 않고 +타사 장비에서 워크로드를 실행하기만 하면 된다. +CPU 사용량, 메모리 및 디스크 요청과 같은 항목에 대한 요금이 부과된다. +- *관리형 컨트롤 플레인*: 쿠버네티스 서비스 공급자가 +클러스터 컨트롤 플레인의 확장 및 가용성을 관리하고 패치 및 업그레이드를 처리하도록 한다. +- *관리형 워커 노드*: 필요에 맞는 노드 풀을 정의하면, +쿠버네티스 서비스 공급자는 해당 노드의 가용성 및 +필요 시 업그레이드 제공을 보장한다. +- *통합*: 쿠버네티스를 스토리지, 컨테이너 레지스트리, +인증 방법 및 개발 도구와 같이 +사용자가 필요로 하는 여러 서비스를 통합 제공하는 업체도 있다. + +프로덕션 쿠버네티스 클러스터를 직접 구축하든 파트너와 협력하든, +요구 사항이 *컨트롤 플레인*, *워커 노드*, +*사용자 접근*, *워크로드 자원*과 관련되기 때문에, +다음 섹션들을 검토하는 것이 바람직하다. + +## 프로덕션 클러스터 구성 + +프로덕션 수준 쿠버네티스 클러스터에서, +컨트롤 플레인은 다양한 방식으로 여러 컴퓨터에 분산될 수 있는 서비스들을 통해 +클러스터를 관리한다. +반면, 각 워커 노드는 쿠버네티스 파드를 실행하도록 구성된 단일 엔티티를 나타낸다. + +### 프로덕션 컨트롤 플레인 + +가장 간단한 쿠버네티스 클러스터는 모든 컨트롤 플레인 및 워커 노드 서비스가 +하나의 머신에 실행되는 클러스터이다. +[쿠버네티스 컴포넌트](/ko/docs/concepts/overview/components/) +그림에 명시된 대로, 워커 노드를 추가하여 해당 환경을 확장할 수 있다. +클러스터를 단기간만 사용하거나, +심각한 문제가 발생한 경우 폐기하는 것이 가능하다면, 이 방식을 선택할 수 있다. + +그러나 더 영구적이고 가용성이 높은 클러스터가 필요한 경우 +컨트롤 플레인 확장을 고려해야 한다. +설계 상, 단일 시스템에서 실행되는 단일 시스템 컨트롤 플레인 서비스는 +가용성이 높지 않다. +클러스터를 계속 유지하면서 문제가 발생한 경우 복구할 수 있는지 여부가 중요한 경우, +다음 사항들을 고려한다. + +- *배포 도구 선택*: kubeadm, kops, kubespray와 같은 도구를 이용해 +컨트롤 플레인을 배포할 수 있다. +[배포 도구로 쿠버네티스 설치하기](/ko/docs/setup/production-environment/tools/)에서 +여러 배포 도구를 이용한 프로덕션 수준 배포에 대한 팁을 확인한다. +배포 시, 다양한 +[컨테이너 런타임](/ko/docs/setup/production-environment/container-runtimes/)을 사용할 수 있다. +- *인증서 관리*: 컨트롤 플레인 서비스 간의 보안 통신은 인증서를 사용하여 구현된다. +인증서는 배포 중에 자동으로 생성되거나, 또는 자체 인증 기관을 사용하여 생성할 수 있다. +[PKI 인증서 및 요구 조건](/ko/docs/setup/best-practices/certificates/)에서 +상세 사항을 확인한다. +- *apiserver를 위한 로드밸런서 구성*: 여러 노드에서 실행되는 apiserver 서비스 인스턴스에 +외부 API 호출을 분산할 수 있도록 로드밸런서를 구성한다. +[외부 로드밸런서 생성하기](/docs/tasks/access-application-cluster/create-external-load-balancer/)에서 +상세 사항을 확인한다. +- *etcd 서비스 분리 및 백업*: etcd 서비스는 +다른 컨트롤 플레인 서비스와 동일한 시스템에서 실행되거나, +또는 추가 보안 및 가용성을 위해 별도의 시스템에서 실행될 수 있다. +etcd는 클러스터 구성 데이터를 저장하므로 +필요한 경우 해당 데이터베이스를 복구할 수 있도록 etcd 데이터베이스를 정기적으로 백업해야 한다. +[etcd FAQ](https://etcd.io/docs/v3.4/faq/)에서 etcd 구성 및 사용 상세를 확인한다. +[쿠버네티스를 위한 etcd 클러스터 운영하기](/docs/tasks/administer-cluster/configure-upgrade-etcd/)와 +[kubeadm을 이용하여 고가용성 etcd 생성하기](/docs/setup/production-environment/tools/kubeadm/setup-ha-etcd-with-kubeadm/)에서 +상세 사항을 확인한다. +- *다중 컨트롤 플레인 시스템 구성*: 고가용성을 위해, +컨트롤 플레인은 단일 머신으로 제한되지 않아야 한다. +컨트롤 플레인 서비스가 init 서비스(예: systemd)에 의해 실행되는 경우, +각 서비스는 최소 3대의 머신에서 실행되어야 한다. +그러나, 컨트롤 플레인 서비스를 쿠버네티스 상의 파드 형태로 실행하면 +각 서비스 복제본 요청이 보장된다. +스케줄러는 내결함성이 있어야 하고, 고가용성은 필요하지 않다. +일부 배포 도구는 쿠버네티스 서비스의 리더 선출을 수행하기 위해 +[Raft](https://raft.github.io/) 합의 알고리즘을 설정한다. +리더를 맡은 서비스가 사라지면 다른 서비스가 스스로 리더가 되어 인계를 받는다. +- *다중 영역(zone)으로 확장*: 클러스터를 항상 사용 가능한 상태로 유지하는 것이 중요하다면 +여러 데이터 센터(클라우드 환경에서는 '영역'이라고 함)에서 실행되는 +클러스터를 만드는 것이 좋다. +영역의 그룹을 지역(region)이라고 한다. +동일한 지역의 여러 영역에 클러스터를 분산하면 +하나의 영역을 사용할 수 없게 된 경우에도 클러스터가 계속 작동할 가능성을 높일 수 있다. +[여러 영역에서 실행](/ko/docs/setup/best-practices/multiple-zones/)에서 상세 사항을 확인한다. +- *구동 중인 기능 관리*: 클러스터를 계속 유지하려면, +상태 및 보안을 유지하기 위해 수행해야 하는 작업이 있다. +예를 들어 kubeadm으로 클러스터를 생성한 경우, +[인증서 관리](/ko/docs/tasks/administer-cluster/kubeadm/kubeadm-certs/)와 +[kubeadm 클러스터 업그레이드하기](/ko/docs/tasks/administer-cluster/kubeadm/kubeadm-upgrade/)에 대해 도움이 되는 가이드가 있다. +[클러스터 운영하기](/ko/docs/tasks/administer-cluster/)에서 +더 많은 쿠버네티스 관리 작업을 볼 수 있다. + +컨트롤 플레인 서비스를 실행할 때 사용 가능한 옵션에 대해 보려면, +[kube-apiserver](/docs/reference/command-line-tools-reference/kube-apiserver/), +[kube-controller-manager](/docs/reference/command-line-tools-reference/kube-controller-manager/), +[kube-scheduler](/docs/reference/command-line-tools-reference/kube-scheduler/)를 참조한다. +고가용성 컨트롤 플레인 예제는 +[고가용성 토폴로지를 위한 옵션](/docs/setup/production-environment/tools/kubeadm/ha-topology/), +[kubeadm을 이용하여 고가용성 클러스터 생성하기](/docs/setup/production-environment/tools/kubeadm/high-availability/), +[쿠버네티스를 위한 etcd 클러스터 운영하기](/docs/tasks/administer-cluster/configure-upgrade-etcd/)를 참조한다. +etcd 백업 계획을 세우려면 +[etcd 클러스터 백업하기](/docs/tasks/administer-cluster/configure-upgrade-etcd/#backing-up-an-etcd-cluster)를 참고한다. + +### 프로덕션 워커 노드 + +프로덕션 수준 워크로드는 복원력이 있어야 하고, +이들이 의존하는 모든 것들(예: CoreDNS)도 복원력이 있어야 한다. +컨트롤 플레인을 자체적으로 관리하든 +클라우드 공급자가 대신 수행하도록 하든 상관없이, +워커 노드(간단히 *노드*라고도 함)를 어떤 방법으로 관리할지 고려해야 한다. + +- *노드 구성하기*: 노드는 물리적 또는 가상 머신일 수 있다. +직접 노드를 만들고 관리하려면 지원되는 운영 체제를 설치한 다음 +적절한 [노드 서비스](/ko/docs/concepts/overview/components/#노드-컴포넌트)를 추가하고 실행한다. +다음을 고려해야 한다. + - 워크로드의 요구 사항 (노드가 적절한 메모리, CPU, 디스크 속도, 저장 용량을 갖도록 구성) + - 일반적인 컴퓨터 시스템이면 되는지, 아니면 GPU, 윈도우 노드, 또는 VM 격리를 필요로 하는 워크로드가 있는지 +- *노드 검증하기*: [노드 구성 검증하기](/ko/docs/setup/best-practices/node-conformance/)에서 +노드가 쿠버네티스 클러스터에 조인(join)에 필요한 요구 사항을 +만족하는지 확인하는 방법을 알아본다. +- *클러스터에 노드 추가하기*: 클러스터를 자체적으로 관리하는 경우, +머신을 준비하고, 클러스터의 apiserver에 이를 수동으로 추가하거나 +또는 머신이 스스로 등록하도록 하여 노드를 추가할 수 있다. +이러한 방식으로 노드를 추가하는 방법을 보려면 [노드](/ko/docs/concepts/architecture/nodes/) 섹션을 확인한다. +- *클러스터에 윈도우 노드 추가하기*: 윈도우 컨테이너로 구현된 워크로드를 +실행할 수 있도록, 쿠버네티스는 윈도우 워커 노드를 지원한다. +[쿠버네티스에서의 윈도우](/ko/docs/setup/production-environment/windows/)에서 상세 사항을 확인한다. +- *노드 스케일링*: 클러스터가 최종적으로 필요로 하게 될 용량만큼 +확장하는 것에 대한 계획이 있어야 한다. +실행해야 하는 파드 및 컨테이너 수에 따라 필요한 노드 수를 판별하려면 +[대형 클러스터에 대한 고려 사항](/ko/docs/setup/best-practices/cluster-large/)을 확인한다. +만약 노드를 직접 관리한다면, 직접 물리적 장비를 구입하고 설치해야 할 수도 있음을 의미한다. +- *노드 자동 스케일링*: 대부분의 클라우드 공급자는 +비정상 노드를 교체하거나 수요에 따라 노드 수를 늘리거나 줄일 수 있도록 +[클러스터 오토스케일러](https://github.com/kubernetes/autoscaler/tree/master/cluster-autoscaler#readme)를 지원한다. +[자주 묻는 질문](https://github.com/kubernetes/autoscaler/blob/master/cluster-autoscaler/FAQ.md)에서 +오토스케일러가 어떻게 동작하는지, +[배치](https://github.com/kubernetes/autoscaler/tree/master/cluster-autoscaler#deployment) 섹션에서 +각 클라우드 공급자별로 어떻게 구현했는지를 확인한다. +온프레미스의 경우, 필요에 따라 새 노드를 가동하도록 +스크립트를 구성할 수 있는 가상화 플랫폼이 있다. +- *노드 헬스 체크 구성*: 중요한 워크로드의 경우, +해당 노드에서 실행 중인 노드와 파드의 상태가 정상인지 확인하고 싶을 것이다. +[Node Problem Detector](/docs/tasks/debug-application-cluster/monitor-node-health/) +데몬을 사용하면 노드가 정상인지 확인할 수 있다. + +## 프로덕션 사용자 관리 + +프로덕션에서는, 클러스터를 한 명 또는 여러 명이 사용하던 모델에서 +수십에서 수백 명이 사용하는 모델로 바꿔야 하는 경우가 발생할 수 있다. +학습 환경 또는 플랫폼 프로토타입에서는 모든 작업에 대한 단일 관리 계정으로도 +충분할 수 있다. 프로덕션에서는 여러 네임스페이스에 대한, 액세스 수준이 +각각 다른 더 많은 계정이 필요하다. + +프로덕션 수준의 클러스터를 사용한다는 것은 +다른 사용자의 액세스를 선택적으로 허용할 방법을 결정하는 것을 의미한다. +특히 클러스터에 액세스를 시도하는 사용자의 신원을 확인(인증, authentication)하고 +요청한 작업을 수행할 권한이 있는지 결정(인가, authorization)하기 위한 +다음과 같은 전략을 선택해야 한다. + +- *인증*: apiserver는 클라이언트 인증서, 전달자 토큰, 인증 프록시 또는 +HTTP 기본 인증을 사용하여 사용자를 인증할 수 있다. +사용자는 인증 방법을 선택하여 사용할 수 있다. +apiserver는 또한 플러그인을 사용하여 +LDAP 또는 Kerberos와 같은 조직의 기존 인증 방법을 활용할 수 있다. +쿠버네티스 사용자를 인증하는 다양한 방법에 대한 설명은 +[인증](/docs/reference/access-authn-authz/authentication/)을 참조한다. +- *인가*: 일반 사용자 인가를 위해, RBAC 와 ABAC 중 하나를 선택하여 사용할 수 있다. [인가 개요](/ko/docs/reference/access-authn-authz/authorization/)에서 사용자 계정과 서비스 어카운트 인가를 위한 여러 가지 모드를 확인할 수 있다. + - *역할 기반 접근 제어* ([RBAC](/docs/reference/access-authn-authz/rbac/)): 인증된 사용자에게 특정 권한 집합을 허용하여 클러스터에 대한 액세스를 할당할 수 있다. 특정 네임스페이스(Role) 또는 전체 클러스터(ClusterRole)에 권한을 할당할 수 있다. 그 뒤에 RoleBindings 및 ClusterRoleBindings를 사용하여 해당 권한을 특정 사용자에게 연결할 수 있다. + - *속성 기반 접근 제어* ([ABAC](/docs/reference/access-authn-authz/abac/)): 클러스터의 리소스 속성을 기반으로 정책을 생성하고 이러한 속성을 기반으로 액세스를 허용하거나 거부할 수 있다. 정책 파일의 각 줄은 버전 관리 속성(apiVersion 및 종류), 그리고 '대상(사용자 또는 그룹)', '리소스 속성', '비 리소스 속성(`/version` 또는 `/apis`)' 및 '읽기 전용'과 일치하는 사양 속성 맵을 식별한다. 자세한 내용은 [예시](/docs/reference/access-authn-authz/abac/#examples)를 참조한다. + +프로덕션 쿠버네티스 클러스터에 인증과 인가를 설정할 때, 다음의 사항을 고려해야 한다. + +- *인가 모드 설정*: 쿠버네티스 API 서버([kube-apiserver](/docs/reference/command-line-tools-reference/kube-apiserver/))를 실행할 때, +*`--authorization-mode`* 플래그를 사용하여 인증 모드를 설정해야 한다. +예를 들어, (*`/etc/kubernetes/manifests`*에 있는) +*`kube-adminserver.yaml`* 파일 안의 플래그를 `Node,RBAC`으로 설정할 수 있다. +이렇게 하여 인증된 요청이 Node 인가와 RBAC 인가를 사용할 수 있게 된다. +- *사용자 인증서와 롤 바인딩 생성(RBAC을 사용하는 경우)*: RBAC 인증을 사용하는 경우, +사용자는 클러스터 CA가 서명한 CSR(CertificateSigningRequest)을 만들 수 있다. +그 뒤에 각 사용자에게 역할 및 ClusterRoles를 바인딩할 수 있다. +자세한 내용은 +[인증서 서명 요청](/docs/reference/access-authn-authz/certificate-signing-requests/)을 참조한다. +- *속성을 포함하는 정책 생성(ABAC을 사용하는 경우)*: ABAC 인증을 사용하는 경우, +속성의 집합으로 정책을 생성하여, 인증된 사용자 또는 그룹이 +특정 리소스(예: 파드), 네임스페이스, 또는 apiGroup에 접근할 수 있도록 한다. +[예시](/docs/reference/access-authn-authz/abac/#examples)에서 +더 많은 정보를 확인한다. +- *어드미션 컨트롤러 도입 고려*: +[웹훅 토큰 인증](/docs/reference/access-authn-authz/authentication/#webhook-token-authentication)은 +API 서버를 통해 들어오는 요청의 인가에 사용할 수 있는 추가적인 방법이다. +웹훅 및 다른 인가 형식을 사용하려면 API 서버에 +[어드미션 컨트롤러](/docs/reference/access-authn-authz/admission-controllers/)를 +추가해야 한다. + +## 워크로드에 자원 제한 걸기 + +프로덕션 워크로드의 요구 사항이 +쿠버네티스 컨트롤 플레인 안팎의 압박을 초래할 수 있다. +워크로드의 요구 사항을 충족하도록 클러스터를 구성할 때 다음 항목을 고려한다. + +- *네임스페이스 제한 설정*: 메모리, CPU와 같은 자원의 네임스페이스 별 쿼터를 설정한다. +[메모리, CPU 와 API 리소스 관리](/ko/docs/tasks/administer-cluster/manage-resources/)에서 +상세 사항을 확인한다. +[계층적 네임스페이스](/blog/2020/08/14/introducing-hierarchical-namespaces/)를 설정하여 +제한을 상속할 수도 있다. +- *DNS 요청에 대한 대비*: 워크로드가 대규모로 확장될 것으로 예상된다면, +DNS 서비스도 확장할 준비가 되어 있어야 한다. +[클러스터의 DNS 서비스 오토스케일링](/docs/tasks/administer-cluster/dns-horizontal-autoscaling/)을 확인한다. +- *추가적인 서비스 어카운트 생성*: 사용자 계정은 *클러스터*에서 사용자가 무엇을 할 수 있는지 결정하는 반면에, +서비스 어카운트는 특정 네임스페이스 내의 파드 접근 권한을 결정한다. +기본적으로, 파드는 자신의 네임스페이스의 기본 서비스 어카운트을 이용한다. +[서비스 어카운트 관리하기](/ko/docs/reference/access-authn-authz/service-accounts-admin/)에서 +새로운 서비스 어카운트을 생성하는 방법을 확인한다. 예를 들어, 다음의 작업을 할 수 있다. + - 파드가 특정 컨테이너 레지스트리에서 이미지를 가져 오는 데 사용할 수 있는 시크릿을 추가한다. [파드를 위한 서비스 어카운트 구성하기](/docs/tasks/configure-pod-container/configure-service-account/)에서 예시를 확인한다. + - 서비스 어카운트에 RBAC 권한을 할당한다. [서비스어카운트 권한](/docs/reference/access-authn-authz/rbac/#service-account-permissions)에서 상세 사항을 확인한다. + +## {{% heading "whatsnext" %}} + +- 프로덕션 쿠버네티스를 직접 구축할지, +아니면 [턴키 클라우드 솔루션](/docs/setup/production-environment/turnkey-solutions/) 또는 +[쿠버네티스 파트너](/partners/)가 제공하는 서비스를 이용할지 결정한다. +- 클러스터를 직접 구축한다면, +[인증서](/ko/docs/setup/best-practices/certificates/)를 어떻게 관리할지, +[etcd](/docs/setup/production-environment/tools/kubeadm/setup-ha-etcd-with-kubeadm/)와 +[API 서버](/ko/docs/setup/production-environment/tools/kubeadm/ha-topology/) +등의 기능에 대한 고가용성을 +어떻게 보장할지를 계획한다. +- 배포 도구로 [kubeadm](/ko/docs/setup/production-environment/tools/kubeadm/), [kops](/ko/docs/setup/production-environment/tools/kops/), [Kubespray](/ko/docs/setup/production-environment/tools/kubespray/) 중 +하나를 선택한다. +- [인증](/docs/reference/access-authn-authz/authentication/) 및 +[인가](/ko/docs/reference/access-authn-authz/authorization/) 방식을 선택하여 +사용자 관리 방법을 구성한다. +- [자원 제한](/ko/docs/tasks/administer-cluster/manage-resources/), +[DNS 오토스케일링](/docs/tasks/administer-cluster/dns-horizontal-autoscaling/), +[서비스 어카운트](/ko/docs/reference/access-authn-authz/service-accounts-admin/)를 설정하여 +애플리케이션 워크로드의 실행에 대비한다. diff --git a/content/ko/docs/setup/production-environment/tools/kubeadm/control-plane-flags.md b/content/ko/docs/setup/production-environment/tools/kubeadm/control-plane-flags.md index 358274d143..d978e7d59f 100644 --- a/content/ko/docs/setup/production-environment/tools/kubeadm/control-plane-flags.md +++ b/content/ko/docs/setup/production-environment/tools/kubeadm/control-plane-flags.md @@ -76,7 +76,11 @@ kind: ClusterConfiguration kubernetesVersion: v1.16.0 scheduler: extraArgs: - bind-address: 0.0.0.0 - config: /home/johndoe/schedconfig.yaml - kubeconfig: /home/johndoe/kubeconfig.yaml + config: /etc/kubernetes/scheduler-config.yaml + extraVolumes: + - name: schedulerconfig + hostPath: /home/johndoe/schedconfig.yaml + mountPath: /etc/kubernetes/scheduler-config.yaml + readOnly: true + pathType: "File" ``` From edb849a7253aa5538b2a57fba883bcefe385f29d Mon Sep 17 00:00:00 2001 From: redcometlpb Date: Thu, 3 Jun 2021 10:52:10 +0900 Subject: [PATCH 09/10] Translate concepts/storage/volume-health-monitoring in Korean --- .../storage/volume-health-monitoring.md | 30 +++++++++++++++++++ 1 file changed, 30 insertions(+) create mode 100644 content/ko/docs/concepts/storage/volume-health-monitoring.md diff --git a/content/ko/docs/concepts/storage/volume-health-monitoring.md b/content/ko/docs/concepts/storage/volume-health-monitoring.md new file mode 100644 index 0000000000..ce149165da --- /dev/null +++ b/content/ko/docs/concepts/storage/volume-health-monitoring.md @@ -0,0 +1,30 @@ +--- +title: 볼륨 헬스 모니터링 +content_type: concept +--- + + + +{{< feature-state for_k8s_version="v1.21" state="alpha" >}} + +{{< glossary_tooltip text="CSI" term_id="csi" >}} 볼륨 헬스 모니터링을 통해 CSI 드라이버는 기본 스토리지 시스템에서 비정상적인 볼륨 상태를 감지하고 이를 {{< glossary_tooltip text="PVC" term_id="persistent-volume-claim" >}} 또는 {{< glossary_tooltip text="파드" term_id="pod" >}}의 이벤트로 보고한다. + + + +## 볼륨 헬스 모니터링 + +쿠버네티스 _볼륨 헬스 모니터링_ 은 쿠버네티스가 CSI(Container Storage Interface)를 구현하는 방법의 일부다. 볼륨 헬스 모니터링 기능은 외부 헬스 모니터 컨트롤러와 {{< glossary_tooltip term_id="kubelet" text="kubelet" >}}, 2가지 컴포넌트로 구현된다. + +CSI 드라이버가 컨트롤러 측의 볼륨 헬스 모니터링 기능을 지원하는 경우, CSI 볼륨에서 비정상적인 볼륨 상태가 감지될 때 관련 {{< glossary_tooltip text="퍼시스턴트볼륨클레임" term_id="persistent-volume-claim" >}}(PersistentVolumeClaim, PVC) 이벤트가 보고된다. + +외부 헬스 모니터 {{< glossary_tooltip text="컨트롤러" term_id="controller" >}}는 노드 장애 이벤트도 감시한다. `enable-node-watcher` 플래그를 true로 설정하여 노드 장애 모니터링을 활성화할 수 있다. 외부 헬스 모니터가 노드 장애 이벤트를 감지하면, 컨트롤러는 이 PVC를 사용하는 파드가 장애 상태인 노드에 있음을 나타내는 이벤트가 PVC에 보고된다고 알린다. + +CSI 드라이버가 노드 측에서 볼륨 헬스 모니터링 기능을 지원하는 경우, CSI 볼륨에서 비정상적인 볼륨 상태가 감지되면 PVC를 사용하는 모든 파드에서 이벤트가 보고된다. + +{{< note >}} +노드 측에서 이 기능을 사용하려면 `CSIVolumeHealth` [기능 게이트](/ko/docs/reference/command-line-tools-reference/feature-gates/)를 활성화해야 한다. +{{< /note >}} + +## {{% heading "whatsnext" %}} + +이 기능을 구현한 CSI 드라이버를 확인하려면 [CSI 드라이버 문서](https://kubernetes-csi.github.io/docs/drivers.html)를 참고한다. From 104ec652828bedc29097f18774f454ca559f965e Mon Sep 17 00:00:00 2001 From: jmkim Date: Wed, 26 May 2021 14:04:35 +0900 Subject: [PATCH 10/10] Translate /concepts/cluster-administration/system-logs.md into Korean --- .../cluster-administration/system-logs.md | 148 ++++++++++++++++++ 1 file changed, 148 insertions(+) create mode 100644 content/ko/docs/concepts/cluster-administration/system-logs.md diff --git a/content/ko/docs/concepts/cluster-administration/system-logs.md b/content/ko/docs/concepts/cluster-administration/system-logs.md new file mode 100644 index 0000000000..13008ebbd8 --- /dev/null +++ b/content/ko/docs/concepts/cluster-administration/system-logs.md @@ -0,0 +1,148 @@ +--- + + + +title: 시스템 로그 +content_type: concept +weight: 60 +--- + + + +시스템 컴포넌트 로그는 클러스터에서 발생하는 이벤트를 기록하며, 이는 디버깅에 아주 유용하다. +더 많거나 적은 세부 정보를 표시하도록 다양하게 로그를 설정할 수 있다. +로그는 컴포넌트 내에서 오류를 표시하는 것 처럼 간단하거나, 이벤트의 단계적 추적(예: HTTP 엑세스 로그, 파드의 상태 변경, 컨트롤러 작업 또는 스케줄러의 결정)을 표시하는 것처럼 세밀할 수 있다. + + + +## Klog + +klog는 쿠버네티스의 로깅 라이브러리다. [klog](https://github.com/kubernetes/klog) +는 쿠버네티스 시스템 컴포넌트의 로그 메시지를 생성한다. + +klog 설정에 대한 더 많은 정보는, [커맨드라인 툴](/docs/reference/command-line-tools-reference/)을 참고한다. + +klog 네이티브 형식 예 : +``` +I1025 00:15:15.525108 1 httplog.go:79] GET /api/v1/namespaces/kube-system/pods/metrics-server-v0.3.1-57c75779f-9p8wg: (1.512ms) 200 [pod_nanny/v0.0.0 (linux/amd64) kubernetes/$Format 10.56.1.19:51756] +``` + +### 구조화된 로깅 + +{{< feature-state for_k8s_version="v1.19" state="alpha" >}} + +{{}} +구조화된 로그메시지로 마이그레이션은 진행중인 작업이다. 이 버전에서는 모든 로그 메시지가 구조화되지 않는다. 로그 파일을 +파싱할 때, 구조화되지 않은 로그 메시지도 처리해야 한다. + +로그 형식 및 값 직렬화는 변경될 수 있다. +{{< /warning>}} + +구조화된 로깅은 로그 메시지에 통일된 구조를 적용하여 정보를 쉽게 추출하고, +로그를 보다 쉽고 저렴하게 저장하고 처리하는 작업이다. +새로운 메시지 형식은 이전 버전과 호환되며 기본적으로 활성화 된다. + +구조화된 로그 형식: + +```ini + "" ="" ="" ... +``` + +예시: + +```ini +I1025 00:15:15.525108 1 controller_utils.go:116] "Pod status updated" pod="kube-system/kubedns" status="ready" +``` + + +### JSON 로그 형식 + +{{< feature-state for_k8s_version="v1.19" state="alpha" >}} + +{{}} + +JSON 출력은 많은 표준 klog 플래그를 지원하지 않는다. 지원하지 않는 klog 플래그 목록은, [커맨드라인 툴](/docs/reference/command-line-tools-reference/)을 참고한다. + +모든 로그가 JSON 형식으로 작성되는 것은 아니다(예: 프로세스 시작 중). 로그를 파싱하려는 경우 +JSON 형식이 아닌 로그 행을 처리할 수 있는지 확인해야 한다. + +필드 이름과 JSON 직렬화는 변경될 수 있다. +{{< /warning >}} + +`--logging-format=json` 플래그는 로그 형식을 klog 기본 형식에서 JSON 형식으로 변경한다. +JSON 로그 형식 예시(보기좋게 출력된 형태): + +```json +{ + "ts": 1580306777.04728, + "v": 4, + "msg": "Pod status updated", + "pod":{ + "name": "nginx-1", + "namespace": "default" + }, + "status": "ready" +} +``` + +특별한 의미가 있는 키: +* `ts` - Unix 시간의 타임스탬프 (필수, 부동 소수점) +* `v` - 자세한 정도 (필수, 정수, 기본 값 0) +* `err` - 오류 문자열 (선택 사항, 문자열) +* `msg` - 메시지 (필수, 문자열) + + +현재 JSON 형식을 지원하는 컴포넌트 목록: +* {{< glossary_tooltip term_id="kube-controller-manager" text="kube-controller-manager" >}} +* {{< glossary_tooltip term_id="kube-apiserver" text="kube-apiserver" >}} +* {{< glossary_tooltip term_id="kube-scheduler" text="kube-scheduler" >}} +* {{< glossary_tooltip term_id="kubelet" text="kubelet" >}} + +### 로그 정리(sanitization) + +{{< feature-state for_k8s_version="v1.20" state="alpha" >}} + +{{}} +로그 정리(sanitization)는 상당한 오버 헤드를 발생시킬 수 있으므로 프로덕션 환경에서는 사용하지 않아야한다. +{{< /warning >}} + + `--experimental-logging-sanitization` 플래그는 klog 정리(sanitization) 필터를 활성화 한다. +활성화된 경우 모든 로그 인자에서 민감한 데이터(예: 비밀번호, 키, 토큰)가 표시된 필드를 검사하고 +이러한 필드의 로깅이 방지된다. + +현재 로그 정리(sanitization)를 지원하는 컴포넌트 목록: +* kube-controller-manager +* kube-apiserver +* kube-scheduler +* kubelet + +{{< note >}} +로그 정리(sanitization) 필터는 사용자 작업 로그로부터 민감한 데이터가 유출되는 것을 방지할 수 없다. +{{< /note >}} + +### 로그 상세 레벨(verbosity) + +`-v` 플래그로 로그 상세 레벨(verbosity)을 제어한다. 값을 늘리면 기록된 이벤트 수가 증가한다. 값을 줄이면 +기록된 이벤트 수가 줄어든다. +로그 상세 레벨(verbosity)를 높이면 점점 덜 심각한 이벤트가 기록된다. 로그 상세 레벨(verbosity)을 0으로 설정하면 중요한 이벤트만 기록된다. + +### 로그 위치 + +시스템 컴포넌트에는 컨테이너에서 실행되는 것과 컨테이너에서 실행되지 않는 두 가지 유형이 있다. 예를 들면 다음과 같다. + +* 쿠버네티스 스케줄러와 kube-proxy는 컨테이너에서 실행된다. +* kubelet과 컨테이너 런타임(예: 도커)은 컨테이너에서 실행되지 않는다. + +systemd를 사용하는 시스템에서는, kubelet과 컨테이너 런타임은 jounald에 기록한다. +그 외 시스템에서는, `/var/log` 디렉터리의 `.log` 파일에 기록한다. +컨테이너 내부의 시스템 컴포넌트들은 기본 로깅 메커니즘을 무시하고, +항상 `/var/log` 디렉터리의 `.log` 파일에 기록한다. +컨테이너 로그와 마찬가지로, `/var/log` 디렉터리의 시스템 컴포넌트 로그들은 로테이트해야 한다. +`kube-up.sh` 스크립트로 생성된 쿠버네티스 클러스터에서는, `logrotate` 도구로 로그가 로테이트되도록 설정된다. +`logrotate` 도구는 로그가 매일 또는 크기가 100MB 보다 클 때 로테이트된다. + +## {{% heading "whatsnext" %}} + +* [쿠버네티스 로깅 아키텍처](/docs/concepts/cluster-administration/logging/) 알아보기 +* [구조화된 로깅](https://github.com/kubernetes/enhancements/tree/master/keps/sig-instrumentation/1602-structured-logging) 알아보기 +* [로깅 심각도(serverity) 규칙](https://github.com/kubernetes/community/blob/master/contributors/devel/sig-instrumentation/logging.md) 알아보기