[[["容易理解","easyToUnderstand","thumb-up"],["確實解決了我的問題","solvedMyProblem","thumb-up"],["其他","otherUp","thumb-up"]],[["難以理解","hardToUnderstand","thumb-down"],["資訊或程式碼範例有誤","incorrectInformationOrSampleCode","thumb-down"],["缺少我需要的資訊/範例","missingTheInformationSamplesINeed","thumb-down"],["翻譯問題","translationIssue","thumb-down"],["其他","otherDown","thumb-down"]],["上次更新時間：2025-07-22 (世界標準時間)。"],[],[],null,["# GKE on AWS architecture\n\nOverview\n--------\n\nGKE on AWS is hybrid cloud software that extends\nGoogle Kubernetes Engine (GKE) to Amazon Web Services (AWS).\n\nGKE on AWS uses regular AWS resources such as\n[Elastic Compute Cloud (EC2)](https://aws.amazon.com/ec2/),\n[Elastic Block Storage (EBS)](https://aws.amazon.com/ebs/),\nand [Elastic Load Balancer (ELB)](https://aws.amazon.com/elasticloadbalancing/?nc2=h_ql_prod_nt_elb).\nMost AWS resources that GKE on AWS creates have names that start\nwith `gke-`.\n\nArchitecture\n------------\n\nThere are two components to GKE on AWS.\n\n1. *The management service*, an environment that can install and update your user clusters, uses the AWS API to provision resources.\n2. *User clusters*, where you run your workloads.\n\nThis topic describes the purpose and shape of your Anthos management service and\nuser clusters.\n\nManagement service\n------------------\n\nThe management service provides and manages components of your\nGKE on AWS installation. For example, you create user clusters\nusing the management service. The management service uses the AWS API to\nprovision resources.\n\nYou can create your management\nservice in a [dedicated AWS VPC](/kubernetes-engine/multi-cloud/docs/aws/previous-generation/how-to/installing-management)\nor an\n[existing AWS VPC](/kubernetes-engine/multi-cloud/docs/aws/previous-generation/how-to/integrating-existing-infrastructure).\n\nYou need a management service in every AWS Virtual Private Cloud (VPC) where you\nrun GKE on AWS. The management service is installed in one AWS\nAvailability Zone. You only need one management service per VPC; a management\nservice can manage multiple user clusters.\n\nThe primary component of the management service is the Cluster Operator. The\nCluster Operator is a\n[Kubernetes Operator](https://kubernetes.io/docs/concepts/extend-kubernetes/operator/)\nthat creates and manages your AWSClusters and\nAWSNodePools. The Cluster Operator stores configuration in an etcd\ndatabase with storage persisted on an AWS EBS volume.\n\n### Installing and configuring your management service\n\nThis section describes the tools you can use to manage your management service.\n\n#### The anthos-gke tool\n\nYou create and manage your clusters with the\n[`anthos-gke`](/kubernetes-engine/multi-cloud/docs/aws/previous-generation/how-to/prerequisites#anthos_gke_command-line_tool) command-line\ntool. For more information, see\n[The anthos-gke tool](/kubernetes-engine/multi-cloud/docs/aws/previous-generation/reference/anthos-gke).\n\n#### Connect\n\nWith [Connect](/anthos/multicluster-management/connect), you can view\nand sign in to your GKE on AWS and GKE clusters on\nGoogle Cloud from the same interface in\n[Google Cloud console](/anthos/multicluster-management/console). All of your\nresources are shown in a single dashboard, so you can get visibility into\nyour workloads across multiple Kubernetes clusters.\n\nUser clusters\n-------------\n\nA user cluster includes two components, both of which are Kubernetes custom\nresources hosted by the management service:\n\n1. A control plane.\n2. One or more AWSNodePools.\n\n### AWSCluster\n\nAn AWSCluster runs in a single VPC.\n\nWhen you install a management cluster into\na [Dedicated VPC](/kubernetes-engine/multi-cloud/docs/aws/previous-generation/how-to/installing-management),\nGKE on AWS creates control plane replicas in every zone you\nspecify in `dedicatedVPC.availabilityZones`. When you install a management\ncluster into\n[existing infrastructure](/kubernetes-engine/multi-cloud/docs/aws/previous-generation/how-to/integrating-existing-infrastructure)\nGKE on AWS creates an AWSCluster\nwith three control plane replicas in the same availability zones.\nEach replica belongs to its own\n[AWS Auto Scaling group](https://docs.aws.amazon.com/autoscaling/ec2/userguide/AutoScalingGroup.html)\nwhich restarts instances when they are terminated.\n\nThe management service places the control planes in a private\nsubnet behind an AWS Network Load Balancer (NLB). The management service\ninteracts with the control plane using NLB.\n\nTo create control planes across multiple AWS availability zones,\nsee [High availability user clusters](/kubernetes-engine/multi-cloud/docs/aws/previous-generation/how-to/high-availability).\n\nEach control plane stores configuration in a local etcd database. These\ndatabases are replicated and set up in a\n[stacked high availability topology](https://kubernetes.io/docs/setup/production-environment/tools/kubeadm/ha-topology/#stacked-etcd-topology).\n\nOne control plane manages one or more AWSNodePools.\n\n### AWSNodePool\n\nAWSNodePools function like GKE\n[Node Pools on Google Cloud](/kubernetes-engine/docs/concepts/node-pools).\nA node pool is a group of nodes within a cluster that all have the same\nconfiguration. Node pools use the AWSNodePool resource and can contain\none or more nodes. Each node pool belongs to its own AWS Auto Scaling group\nwhich restarts instances when they are terminated.\n\nTroubleshooting\n---------------\n\nYou can troubleshoot your GKE on AWS installation by viewing\nKubernetes Events from your AWSCluster and AWSNodePools. For\nmore information, see the\n[Troubleshooting](/kubernetes-engine/multi-cloud/docs/aws/previous-generation/troubleshooting) guide.\n\nWhat's next\n-----------\n\n- [Install a management service](/kubernetes-engine/multi-cloud/docs/aws/previous-generation/how-to/installing-management) in a dedicated VPC with a bastion host.\n- [Integrating existing infrastructure](/kubernetes-engine/multi-cloud/docs/aws/previous-generation/how-to/integrating-existing-infrastructure) in an existing VPC."]]
