The Four Pillars of Observability
summary
CSV upload API shows P95 latency slowly creeping up; metrics (CPU, memory, DB, PHP‑FPM) look normal but logs only record CSV import finished messages; tracing reveals a 16s root span with a ~15s custom CSV span and only ~5% DB time, pointing to PHP work; continuous profiling finds 495,000 exceptions in 14.2s (~35k/sec) costing ~23% of runtime caused by PHPUnit comparator exceptions from unexpected input data; solution is concrete: trace to span, correlate logs, sample profile to find and fix the failing comparisons.
AI-written summary · about
chapters (9)
- 0:00Introduction
- 1:03Customer case study
- 2:05First pillar: Metrics types
- 4:11Metrics solving real-world problems
- 8:23Second pillar: Logs structure
- 13:26Third pillar: Tracing spans
- 18:23Fourth pillar: Profiling types
- 24:24Beyond big failures
- 25:48Connecting the pillars
related talks
44mMonitor Your PHP Applications with Logs, Metrics, Pings, and Traces
19mYou can't test everything, but you should monitor it
39mConnecting the dots: observability in large e-commerce environments
01h 04mProfiling PHP Applications
48mFantastic Performance and Where to Find It
47mResolving performance issues in practice
32mTackling Slow Queries: A Practical Approach to Prevention and Correction
43mOpenTelemetry with PHP is easier than you think!