Java Performance Analysis
summary
Live demos diagnose real incidents: start from an SLA violation, use vmstat to split user vs kernel time, spot high iowait and context switches indicating block I/O, and build a time budget that shows the database consumes ~200ms of a 300ms budget. Capture low-overhead recordings with Java Flight Recorder (JFR), use VisualVM/generational accounting to spot a leak by high generation counts, and fix a DecaPo regression by enlarging G1 region size to eliminate humongous allocations due to 1MB regions, cutting GC pause overhead (about 40–50%).
AI-written summary · about
chapters (8)
- 0:00Performance tuning methodology
- 3:24Analyzing system metrics
- 11:01Time budgeting techniques
- 13:26Latency in microservices
- 20:38Garbage collection analysis
- 26:33Understanding G1 GC mechanics
- 35:42Profiling memory leaks
- 43:47Advanced diagnostic tools
related talks
01h 28mThe dark art of performance tuning or how to become a performance hero ...
47mJVM Performance Engineering - Inside OpenJDK and the HotSpot Java Virtual Machine
38mMemory Efficient Java
21mMemory Efficient Java
33mChecking out JVM 12-16
42mOff-CPU Profiling & Latency Diagnostics in Java
51mThe Power of JDK Flight Recorder: Efficient Profiling and Troubleshooting for Java Applications
45mUnder the Hood of a JVM Application - Java Flight Recorder / Java Mission Control