How to Effectively Spy on Your Systems
summary
OpenTelemetry becomes the standard for traces logs and metrics and the talk shows how to use it end to end: instrument edges first with existing libraries, correlate signals via TraceID/spanID and baggage API, and avoid PII and cardinality explosions (customer IDs in metric labels). Use processors to redact/enrich, choose head vs tail-based sampling to trade cost for fidelity, and deploy an OpenTelemetry Collector (sidecar or gateway) to centralize sampling and reduce per-service risk.
AI-written summary · about
chapters (21)
- 0:00Introduction
- 1:48Observability matters
- 6:23OpenTelemetry role
- 8:29Logs challenges and improvements
- 13:05Metrics and traces in OTel
- 15:33Logs vs metrics vs traces
- 19:05Start at system edges
- 22:35Observability strategy
- 27:52Observability best practices
- 30:25Failure scenarios and sensitive data
- 33:56Baggage API for context
- 36:27Avoid sensitive data
- 39:15Code reviews for observability
- 41:02Telemetry processors
- 42:04Correlate logs metrics traces
- 44:14Balance telemetry and cardinality
- 46:02Trace sampling strategies
- 54:11Direct export challenges
- 54:52OTel collector deployment
- 59:04Scaling observability
- 1:00:31Observability driven development
related talks
52mHow to effectively spy on your systems
36mObservability For You and Me with OpenTelemetry
15mOpenTelemetry: The Vision, the Reality and How To Get Started
20mObservability and Open Telemetry
02h 41mThe Science of Signals: Mastering Telemetry for Observability
32mOpenTelemetry feat. logs, metrics and traces
42mFirst Steps with Observability: OpenTelemetry and You
21mOpen Telemetry