The Lean Observability Stack: Quick and Native Telemetry for Service...
summary
When timelines and air‑gaps block full observability, Istio+Envoy become the primary telemetry sources: metrics like Istio requests total, request duration and bytes with rich source/destination labels feed Prometheus and Grafana, SLOs and alerts work without app changes, and access logs stream via OTel collector to Loki. Day‑2 fixes use Prometheus relabeling to cut cardinality (we hit ~50k series), custom telemetry labels add per‑service detail, and autoscaling uses the Prometheus adapter.
AI-written summary · about
chapters (12)
- 0:00Lean observability with service mesh
- 1:24Istio sidecars capture telemetry
- 3:54Prometheus scraping pipeline
- 7:02Istio metrics alerts dashboards
- 8:49Customizing Istio metrics
- 10:56High cardinality in Envoy metrics
- 13:22Mitigating metric explosion
- 15:10Scaling Kubernetes with Istio
- 16:55OpenTelemetry proxy logs
- 19:44Debugging with Envoy logger
- 21:50Native telemetry benefits
- 22:33Q&A on pod errors
related talks
42mOpen Source for Better Observability
13mArchitecting an open observability stack
24mObservability for Istio Service Mesh
27mObservability with Prometheus and beyond
12mThe Road Ahead: Istio Observability
46mObservability Ecosystem in Kubernetes
36mIstio: Real world scenarios
28mMonitoring Stack was yesterday – Observability Platform at scale!