Continuous Profiling in the Cloud-Native era
summary
Parca collects pprof profiles from Go and other runtimes and also uses an eBPF-based Parca agent that samples stacks about 19 samples/sec in-kernel and reads aggregated maps every 10 seconds, stores labeled profiles in object storage with build IDs, and shows flame graphs and source view. It compares pre/post deployments, supports zero-instrumentation for many Python and Ruby versions, and claims organizations often recover up to 30% of CPU by acting on the continuous profiles.
AI-written summary · about
chapters (15)
- 0:00Introduction
- 1:48What is profiling
- 3:15Why profiling matters
- 8:30Challenges of manual profiling
- 9:34Need for continuous profiling
- 12:03How continuous profiling works
- 15:53eBPF and Parker agent
- 20:50Profiles with metadata
- 23:39Zero-instrumentation for Python/Ruby
- 26:29Flame graphs for profiles
- 29:01Source view and demos
- 30:28Improving programs with profiling
- 31:32Storage and querying profiles
- 34:24Compare profiles before/after
- 35:08Polar Signals Cloud and Parker
related talks
36mContinuous Profiling with Parca- Beneath the Kernel to Beyond...
28mConprof - Profiling in the cloud-native era
36mProfiling in the cloud-native era - Matthias Loibl
56mContinuous Profiling the missing piece in your observability puzzle
01h 13mContinuous Profiling: The Missing Piece in Your Observability Puzzle
40mContinuous Profiling, the missing piece in your observability puzzle!
24mCommunity Driven Breakout Session: Continuous Profiling
08mStreamlining continuous profiling with snaps and charms!