Resilient Architectures on Amazon EKS: High Availability Patterns and Strategies
summary
Amazon EKS provides a highly available control plane while worker nodes can run on Fargate, managed node groups, or self-managed nodes; true resilience is implemented at the application layer. Use Kubernetes pod distribution across Availability Zones with pod anti-affinity or TopologySpreadConstraints (k8s 1.19) and preferred scheduling, plus proper startup, liveness, and readiness probes (examples: 5-minute startup, HTTP/TCP/exec). Manage deployments with rollback and monitoring, reduce release risk with canary strategies via Argo Rollouts or Flux (blue/green is also possible but doubles resource cost), and automate rollbacks/metrics because many deploys cause incidents.
AI-written summary · about
chapters (8)
- 0:00Welcome
- 3:35Resilient systems
- 7:04Amazon EKS architecture
- 9:57Pod distribution for HA
- 16:43Kubernetes health checks
- 20:58Deployment strategies for downtime avoidance
- 26:40Resilience vs antifragility
- 29:06Resources and closing
