A service mesh allows to standardize monitoring and observability of applications, to improve reliability.

Kuma is our current candidate, it uses Envoy as TCP/HTTP proxy, play nice with Kubernetes, and offer integrations with Prometheus (T1623) and Elastic stack (T1624).

There are two main goals in this:

• Replace nginx as proxy to access services hosted in containers
• Collect logs and metrics for those services

Plan is to so enable Kuma on Kubernetes (once T1622 is done) and enable relevant policies for monitoring like log metrics and tracing.