Discovering the Magic Behind OpenTelemetry Instrumentation
summary
Argues manual instrumentation gives richer, actionable context than relying solely on automatic hooks (monkey patching, JVM bytecode, eBPF), and demonstrates a client/server demo that uses 3–5 metrics to produce success rate and per-method latency histograms. Shows pipeline: API/SDK instrumentation -> OpenTelemetry Collector -> VictoriaMetrics (metrics), Victoria Logs, Victoria Traces -> Grafana, and warns that automatic instrumentation can flood backends and hurt performance. Emphasizes semantic conventions and context propagation, and flags span events deprecated so correlate logs with trace IDs or use collector-generated metrics.
AI-written summary · about
chapters (18)
- 0:00Introduction
- 1:45What OpenTelemetry is
- 3:08Demystifying instrumentation
- 5:56Why manual instrumentation
- 7:19Instrumenting metrics
- 9:48API vs SDK and vendor neutrality
- 11:13Live demo setup
- 13:24Logs and observability
- 15:29Enriching logs with Collector
- 18:01Logging drawbacks
- 19:06Why tracing is essential
- 20:30Understanding traces and spans
- 22:11Semantic conventions
- 23:15Context propagation
- 24:16Visualizing traces
- 25:39Metrics from traces
- 27:26Recap and key takeaways
- 28:29Q&A
related talks
06mIgnite | Discovering the Magic Behind OpenTelemetry Instrumentation
23mModern Application Debugging: An Intro to OpenTelemetry
22mFine-Tuning Auto-Instrumentation
38mPractical introduction to OpenTelemetry tracing
02h 02mHow To Troubleshoot and Monitor Applications using OpenTelemetry
01h 28mUnderstanding Distributed Systems with OpenTelemetry
36mObservability For You and Me with OpenTelemetry
50mBootiful OpenTelemetry