Question

A platform team runs dozens of microservices and wants uniform retries, traffic splitting for canary releases, and per-service latency metrics — all without modifying or rebuilding the application code. Which approach best achieves this?

Accepted Answer

Deploy a service mesh that injects a sidecar proxy into each pod to handle traffic and telemetry

Answer

Add a NetworkPolicy to each namespace to control traffic flow and gather metrics

Answer

Configure liveness and readiness probes on every Deployment to manage retries and latency

Answer

Use a LoadBalancer Service per microservice so the cloud provider handles retries and canaries