resham_acharya.exe000%
Compiling experience...
All Essays
DevOpsKubernetesJan 18, 202611 min read

Most teams use 12% of Kubernetes to solve 100% of their problems. Pick the slice you actually need and ignore the rest until you can explain why you can't.

I keep meeting teams running Argo, Istio, Knative, OPA, and a service mesh on top of fifteen pods. I keep meeting other teams running plain Deployments and a HorizontalPodAutoscaler and shipping faster.

The 12% you almost certainly need

  1. 01Deployments + ReplicaSets
  2. 02Services + Ingress
  3. 03ConfigMaps + Secrets (sealed via your KMS, please)
  4. 04PersistentVolumeClaims when you have to (you usually don't)
  5. 05HorizontalPodAutoscaler scaled by the metric that actually predicts saturation

That's it. That's the post. Add the next thing only when you can articulate the failure mode it's solving for.

Complexity is a loan. The interest compounds during incidents.