A Profile of Performance Profiling With pprof
summary
Uses pprof with gperftools and language wrappers (perftools.rb, stackprof, Go builtin) to turn samples into actionable hotspots. Shows ld_preload linking on Linux and linker flags on macOS, and Rust hiding symbols unless linked specially. Reports concrete findings: 4140 samples with 51.2% time in GC on a Rails app, alive_neighbors consuming 62.9% and inject 55.4% inside the block, and a real case that drops page load from ~20s to under 3s after memoizing one call.
AI-written summary · about
chapters (9)
- 0:00Overview
- 2:10How to Generate a Profile
- 6:07Profile Output
- 11:11Profile in Rust
- 15:01Profile in Ruby
- 16:41Profile in pprof
- 22:41Live Neighbors
- 30:17pprof Tools
- 36:19Outro
related talks
25mFrom Continuous Profiling to Continuous Optimization
48mProfiling is for Dev, Monitoring is for Ops – no more!
01h 13mContinuous Profiling: The Missing Piece in Your Observability Puzzle
06mOptimizing Performance - Continuous Production Profiling
26mOptimize Performance with Continuous Production Profiling
53mProfiling and Debugging .NET Core Apps on Linux.
01h 01mDebugging and Profiling .NET Core Apps on Linux
46mProfiling your Java Application A Beginner’s Guide