Practical Performance Analysis
summary
Covers a hands‑on method to measure and improve web server performance with concrete tooling and numbers: set up loaders, a probe and OS/JVM tuning, run steady-state vs limit load testing to find the real bottleneck (clients often fail before the server), enable full resource tracking on clients and server, profile steady state with an async profiler to generate flamegraphs, and apply the USE method (errors → saturation → utilization) iteratively until a measurable CPU/network/GC bottleneck is fixed.
AI-written summary · about
chapters (12)
- 0:00Performance tuning intro
- 1:56Performance goals and load testing
- 8:39Load testing environment setup
- 18:08Resource tracking for measurements
- 21:42Baselines and load tests
- 22:45USE method analysis
- 27:24Flame graph profiling
- 29:55Jetty performance war stories
- 34:51Threading model impact
- 41:56Code optimization with flame graphs
- 43:20API design for performance
- 45:56Key takeaways
related talks
38mPractical Performance Analysis
21mWeb-performance practical tools
44mPerformance Testing: the way to make a true load
42mTuning and Optimizing High-Load J2EE Web Applications
16mFrom Beginner to Expert in Web Performance
57mSite-Speed That Sticks
38mA Practical Look at Performance Theory
47mEvery Millisecond Counts