Source: https://medium.engineering/kubernetes-infrastructure-at-medium-d9e2444932ef

Why?

  • Not reinvent the wheel
  • scaling
  • bin-packing
  • self-healing
  • rollout
  • rollbacks

How?

  • production infrastructure is spread across four AZs in four unique k8s clusters.
  • the ability to ship traffic across AZs when needed via some in-house tooling.
  • Gradual rollout of infrastructure changes across production
  • choose Istio as the service mesh.\

Configuration and Management