Profiling in Production
summary
Talk explains why performance matters and how Odnoklassniki built profilers for production. Covers practical steps—metrics/graphs, local profiling, tests, simple production profiling—and then deeper investigation that led to creating Systrace: how Systrace works, how to implement a similar profiler, and how and why to run Systrace in production.
AI-written summary · about
chapters (16)
- 0:00Introduction to profiling
- 1:07Why performance matters
- 3:55Standard profiling approaches
- 7:28Local profiling limits
- 10:16Production measurements
- 17:05Profiling in production idea
- 22:49How Systrace works
- 26:47Systrace access issues
- 28:32Building a custom tracer
- 33:29Gaps in profiling issues
- 35:17Using Facebook Profiler
- 41:42Production profiling results
- 44:57Profiling in production benefits
- 47:24Comparison with Firebase Performance
- 51:00Q&A part 1
- 1:01:15Q&A part 2
