Skip to main content
Account
Watch Rawkode Live
Tutorial

Hands-on Introduction to Parca

Date
Runtime
1:25:42
Share

About this video

What You'll Learn

  1. How Parca scrapes pprof endpoints to ingest continuous profiles from applications.
  2. How to compare two profiles and spot regressions across application versions.
  3. How flame graphs and icicle graphs reveal hot paths in profiles.

Frederic Branczyk of Polar Signals joins to install Parca on a fresh Kubernetes cluster, walk through the eBPF agent and server, and explore flame graphs, profile comparisons, and pprof scraping for continuous profiling at roughly one percent overhead.

Jump to a chapter

View 53 chapters
  1. 0:00 Intro
  2. 0:48 Introduction
  3. 1:28 Meet Frederick
  4. 1:42 Guest Introduction & Background (Frederic Brancic)
  5. 4:20 What is continuous profiling
  6. 4:25 What is Parca? & Continuous Profiling Explained
  7. 6:30 Benefits of Continuous Profiling (Cost Savings, Performance)
  8. 7:18 Profiling Overhead & Sampling
  9. 10:25 Running Parca in production
  10. 10:45 Running Parca in Production
  11. 12:23 Parca Components (Server & Agent)
  12. 14:15 Why Parca
  13. 16:15 Audience Question
  14. 17:37 Hands-on Demo: Installing Parca on Kubernetes
  15. 18:45 Parca Agent
  16. 21:14 Accessing the Parca UI (Port Forward)
  17. 21:25 Port Forward
  18. 22:05 Dark Mode
  19. 23:00 Exploring the Parca UI
  20. 24:19 Viewing Profile Time Series Data
  21. 25:47 Filtering Profiles by Labels
  22. 26:05 Understanding the Flame/Icicle Graph
  23. 26:25 Compare
  24. 26:51 Comparing Profiles Feature
  25. 28:32 Merging Profiles (Holistic View)
  26. 30:25 Compare Versions
  27. 31:01 Comparing Different Application Versions
  28. 32:05 Flame Graph 101
  29. 34:23 Other Use Cases: Latency & Incident Response
  30. 37:54 Audience Q&A (Metrics, eBPF, Compare)
  31. 37:55 Does Parca profile everything
  32. 39:25 Can Parca compare
  33. 40:15 Do we only get CPU approved fails
  34. 40:18 Audience Q&A (Other Profile Types, pprof)
  35. 41:35 Ingest data into Parca
  36. 42:50 Embedded entrepreneur
  37. 42:51 Demo: Scraping from Go pprof Endpoints
  38. 46:10 Scoping
  39. 47:35 Runtime
  40. 48:25 Memory Usage
  41. 50:30 TLDR
  42. 50:31 Explanation of Different Go Profile Types
  43. 55:00 Autocomplete
  44. 55:20 Parca UI Feature: Query Autocomplete
  45. 57:30 Microservices
  46. 1:07:37 Debugging Microservice Profiles (Symbolization Issue)
  47. 1:10:48 Understanding Symbolization & Debugging Info
  48. 1:13:18 Kernel vs User Space Profiling (eBPF Benefits)
  49. 1:16:14 DWARF Debugging Information & Compile Flags
  50. 1:18:21 Summary of Parca's Value
  51. 1:19:44 The Future of Parca (Storage & Scaling)
  52. 1:21:45 Future Integrations (Autoscaling, PGO)
  53. 1:24:35 Conclusion & Call to Action

Technologies featured

Parca Kubernetes eBPF

Meet the Cast

HOST

David Flanagan

@rawkode

GUEST

Frederic Branczyk

@brancz

Comments and resources

More from Rawkode Live

View all 175 episodes
--:--

Hands-on Introduction to Swamp Club

1:18:47

Hands-on Introduction to Kueue

1:16:54

Hands-on Introduction to Odin

1:30:01

Hands-on Introduction to Iroh

1:15:06

Hands-on Introduction to Yoke

1:31:50

Hands-on Introduction to sympozium

Rawkode Academy newsletter

Academy publications by email

Get an email when Rawkode Academy publishes new tutorials, articles, and videos.

Loading email signup. If it does not load, manage email preferences in account settings.