How to Properly Blame Things for Causing Latency
summary
Compares logging, metrics and tracing with concrete tradeoffs, showing Java Spring Boot + Sleuth and JavaScript (fetch) demos that send spans to Zipkin and visualize a dependency graph, explains instrumentation via the brave library and mentions OpenCensus/OpenTracing, demonstrates header propagation using B3 headers, warns about async and thread-pool propagation bugs, cites typical practices like 1% sampling and trace retention measured in days, and shows how traces quickly reveal causal latency vs noisy logs or aggregated metrics.
AI-written summary · about
chapters (19)
- 0:00Intro
- 2:24Logging Metrics and Tracing
- 5:00Response Time
- 7:13Traces
- 8:56Logs
- 9:47Distributed Tracing
- 10:41What is Instrumentation
- 12:28What is Span
- 13:19What is Tracing
- 16:00Using Operations Specific Identifiers
- 17:17Instrumentation
- 19:10How it works
- 23:53Zipkin
- 24:42Architecture
- 26:10MySQL
- 27:19Java Tracing
- 30:18Demo
- 39:16Scope State
- 42:43Conclusion
related talks
45mLatency tracing in distributed Java applications
01h 11mDistributed Tracing : Latency Analysis for Your Microservices
38mTechniques and Tools for a Coherent Discussion about Performance in Complex...
42mAn introduction to distributed tracing and Zipkin
25mTracing Go
33mDistributed Tracing - Implementing Latency Analysis for microservices on Cloud Foundry
16mPragmatic execution tracing
48mObservability of Your Application