Preemptive monitoring applications in production
summary
A sidecar process uses DiagnosticClient + EventPipe to watch live counters and trigger a memory snapshot, then opens that snapshot with CLRMD with LINQ queries to find leaks, largest objects, duplicate strings and root paths. Snapshots run in about 150–200 ms on Windows (longer on first capture and on Linux), the demo hits ~1000 RPS and finds static dictionaries holding megabytes; the OSS tool includes WPF and web front ends and remote Raspberry Pi testing.
AI-written summary · about
chapters (12)
- 0:00Intro
- 9:32Telemetry vs metrics
- 15:02Production challenges
- 18:43How can I know
- 20:33Challenge 1 Net Client
- 22:41Challenge 2 Memory query
- 27:33Demo
- 32:37Test application
- 35:45SOS
- 39:46Threads
- 47:01Raspberry Pi
- 50:41Conclusion
related talks
55mPower your .NET application with the new generation of diagnostics
43mSelf-Aware Applications: Automatic Production Monitoring
53mDebugging .NET Apps
01h 04mAdvanced .NET debugging techniques from real world investigations
38mEffective App Monitoring in Production
01h 13mContinuous Profiling: The Missing Piece in Your Observability Puzzle
52mApplication Insights: The Tool You Never Knew You Needed
01h 02mThe performance loop—A practical guide to profiling and benchmarking