Protecting the Kubernetes API from Observability Tools Overload

This conference talk from Grafana Labs presenters Mario Macías and Terra Tauri shares a cautionary tale about overloading the Kubernetes API when implementing observability solutions. Learn how the team accidentally took down their largest production cluster by deploying tools that made excessive API calls. Discover the practical techniques they developed to protect the Kubernetes API while maintaining full observability functionality, including configuration changes and service implementations that act as shields between information-hungry tools and the API itself. The presentation explains how to properly integrate pod and service metadata into network and request telemetry for creating effective service graphs without compromising cluster stability.