Services

Resources

Company

Amazon EKS (Elastic Kubernetes Service)

Amazon EKS (Elastic Kubernetes Service) is AWS's managed Kubernetes. AWS runs and scales the control plane, the API server and etcd, across three Availability Zones. You choose how worker nodes run: managed node groups, self-managed EC2, Fargate, Karpenter, or EKS Auto Mode, where AWS also manages nodes.

How it works

The responsibility line matters. AWS patches the control plane. You upgrade it when you choose, then upgrade add-ons such as CoreDNS, kube-proxy and the VPC CNI, then node groups, one minor Kubernetes version at a time.

The control plane costs $0.10 per cluster per hour, about $73 a month, on standard version support. A cluster left on a version past standard support moves to extended support at $0.60 per hour, about $438 a month, so a skipped upgrade has a direct cost.

Why it matters in production

Most EKS problems sit on your side of the line: version skew between add-ons and the control plane, IP exhaustion in VPC subnets, node groups that cannot drain because of Pod Disruption Budgets, and IAM roles for pods. Rehearse upgrades on a staging cluster and read the deprecated API list before each one.

See also

On the One2N blog

References

  1. Amazon EKS User Guide

  2. Amazon EKS pricing

Comic about Amazon EKS: AWS runs the control plane of API servers and etcd across three zones, while the team runs the nodes, version upgrades and add-ons like CNI, CoreDNS and kube-proxy; the control plane costs $73 a month on standard support and $438 on extended support for an old version; Asha upgrades 1.30 to 1.31 in order, control plane, add-ons, then node groups, never skipping a version.