Deep Dive Into Profilers: How CPU Profilers Measure Your Application's Performance
Talkat KubeCon + CloudNativeCon + Open Source Summit China 2021 · on Thursday, 2021-12-09· at 12:10· Open Source Summit 演讲厅, 虚拟会议 (Virtual)·
summary
Covers how CPU profilers (Linux perf, Go pprof, Rust pprof-rs) are commonly used to report function-level CPU time and proposes implementing a task-level CPU profiler that attributes CPU consumption to individual requests. Explains the profiler design differences between traditional application profiling and task profiling, and how task profiling helps measure resource usage in multi-tenant services.
AI-written summary · about
chapters (13)
- 0:00Start
- 0:07About me
- 1:07A Story ...
- 2:48CPU Profiler Usage: Perf
- 6:03How CPU Profiler Works: Sampling
- 7:50How CPU Profiler works: Accumulation
- 8:11CPU Profiler: Trade-off
- 9:45Real-World CPU Profiler Implementation
- 10:13The Problem ...
- 12:09The Problem -- Demo
- 13:24CPU Profiler for Request
- 24:04Continuous Profiling
- 25:27Measure Distributed Applications
related talks
51mProduction Profiling: What, Why and How
44mGo Profiling and Observability from Scratch
24mCommunity Driven Breakout Session: Continuous Profiling
34mTrace-Profiling: A New Way About How to Track Application Behavior
33mBuilding a Go Profiler Using Go
48mProfiling is for Dev, Monitoring is for Ops – no more!
26mProfiling Mastery: Decode Your Data and Elevate Observability
38mFighting latency: the CPU profiler is not your ally