Continuous Profiling with Parca- Beneath the Kernel to Beyond...
summary
Parca deploys an eBPF agent as a DaemonSet that samples stacks at 19 Hz and uploads compressed profiles every 10 seconds to a server backed by FrostDB and object storage, claiming low overhead and zero instrumentation. It discovers systemwide targets (containers or bare metal), resolves ELF/DWARF/JIT symbols for C/C++/Go/Rust/Java/Node/Python/Ruby, offers icicle/flame and before‑/after diffs, and highlights surprising hotspots so teams can cut cloud spend and regressions.
AI-written summary · about
chapters (12)
- 0:00Introduction
- 2:51What is profiling
- 4:57Profile production apps
- 7:24How continuous profiling works
- 10:35Introducing Parca
- 12:42Demo: Parca profile explorer
- 18:27Parca agent data collection
- 23:07Binaries to stack traces
- 25:54Sampling frequency and overhead
- 29:29Supported compilers and runtimes
- 33:02Kernel compatibility
- 34:30Roadmap and community
related talks
36mContinuous Profiling in the Cloud-Native era
36mProfiling in the cloud-native era - Matthias Loibl
36mDeveloping Continuous eBPF Profiler to look Beneath the Kernel to Beyond the Clouds
28mConprof - Profiling in the cloud-native era
26mProfiling Mastery: Decode Your Data and Elevate Observability
01h 13mContinuous Profiling: The Missing Piece in Your Observability Puzzle
08mStreamlining continuous profiling with snaps and charms!
56mContinuous Profiling the missing piece in your observability puzzle