There’s No Free Lunch but Can I Have a Free Sandwich With Observability?
summary
A full observability stack uses Linkerd plus Nginx Ingress Controller, Prometheus, Jaeger, Loki (via promtail) and Grafana to extract metrics, logs and traces from an unmodified HTTP/gRPC to‑do app; Linkerd proxies emit metrics and traces, promtail regexes structure plain text logs, and Grafana combines the data. The free‑sandwich result works: p50 is ~3 ms while p95/p99 spike around 3 s, traces narrow slow requests to the to‑do‑manager and an expensive DB call; instrumenting code with OpenTelemetry confirms the hotspot.
AI-written summary · about
chapters (14)
- 0:00Introduction
- 0:19Overview
- 1:29What is Observability
- 3:38Observability Stack
- 6:10Observability Tools
- 8:38Service Mesh
- 10:05Application
- 11:10Recipe
- 14:02Components
- 15:37Example
- 17:10History
- 18:10Logs
- 20:08Graph
- 25:47Summary
related talks
13mArchitecting an open observability stack
56mFull-stack observability and application monitoring with
42mOpen Source for Better Observability
01h 02mDesigning an open-source observability stack for Kubernetes
25mThe Lean Observability Stack: Quick and Native Telemetry for Service...
28mImprove Monitoring and Observability for Kubernetes with OSS Tools
52mIs your JVM app flying blind? Unmask hidden issues with Observability superpowers!
55mCloud-Native Observability with OpenTelemetry