Amazon Elastic Kubernetes Service

Amazon Elastic Kubernetes Service is AWS's managed Kubernetes service for container orchestration on AWS, on premises and at the edge, billed per cluster hour.

Amazon Elastic Kubernetes Service runs the Kubernetes control plane for its users, so teams deploy containerized applications without installing, securing or scaling that layer themselves. Amazon Web Services, Amazon's cloud business, offers it in its Containers line. The service is certified Kubernetes-conformant, so it exposes the standard Kubernetes API that existing manifests and tooling expect. Clusters run in AWS Regions, on AWS Outposts, on a company's own servers joined as Hybrid Nodes, or through EKS Anywhere. EKS Auto Mode manages compute, storage and networking for a cluster, creating nodes when pods need them and consolidating them as demand falls. Teams that manage their own nodes scale them with Karpenter or the Cluster Autoscaler, and automatic node repair replaces nodes that fail health checks. The Horizontal Pod Autoscaler changes replica counts on resource or custom metrics. CoreDNS gives services stable names, and the AWS Load Balancer Controller provisions Application and Network Load Balancers for Services and Ingress. The Amazon EBS CSI driver creates persistent volumes on demand, and the Secrets and Configuration Provider mounts Secrets Manager and Parameter Store values into pods as files. EKS Capabilities add a managed Argo CD, AWS Controllers for Kubernetes and kro to a cluster. Platform and DevOps teams that already build on AWS use EKS to get upstream Kubernetes without running its control plane. They choose the compute per workload: EC2 instances in managed node groups, Auto Mode, or AWS Fargate pods. Companies with data centres or factory sites bring those machines under the same clusters as their cloud workloads through Hybrid Nodes. Teams that want control plane capacity set ahead of load move a cluster to a Provisioned Control Plane tier. Release pipelines commit manifests to Git and let the Argo CD Capability reconcile each cluster to them. With ACK, developers declare AWS resources as Kubernetes objects next to the applications that use them. IAM governs who may reach each cluster, and CloudWatch collects its logs and metrics for the operations team. When a new Kubernetes version ships, clusters move to it in place, and extended support keeps an older version running while a team prepares the upgrade.