Keeping Track of your deadlines in time critical systems
summary
Claims a lightweight C++ monitoring API that detects deadline violations at runtime across threads, using a background high‑priority monitor plus per‑thread thread‑local state and a stack of nested deadlines. Implementation uses 64‑bit unsigned nanosecond ticks, atomic compare_exchange to invalidate deadlines and keeps the happy path lock‑free on the regular path. Measured overhead is about 100–300 ns per check, scales to many threads and detects wrap‑around if configured.
AI-written summary · about
chapters (24)
- 0:00Introduction
- 2:27What are realtime systems
- 3:51Deadlines
- 6:29What would we do
- 8:10Minimum
- 9:54Multiswell
- 10:52How it works
- 13:29Functionality
- 15:53Performance
- 17:26Limitations
- 21:55Timepoints
- 36:54Overflow tolerance
- 37:13Chronostatic clock
- 39:17Unsigned time
- 40:47Violation check
- 42:42Stack of deadlines
- 45:02Wrap up
- 50:00Runtime variants
- 51:37Cost
- 52:34Scaling
- 54:21Framework
- 55:44Key takeaways
- 57:32Extensions
- 58:53Summary
related talks
01h 06mHow to debug Linux multi threaded code
56mDefending against CDD: Chaos-Driven Delivery
34mMonitoring for devs. The necessity or self-indulgence
42mWhen the OS gets in the way
36mIntroducing a New Performance Monitoring Daemon
42mEffective App Monitoring in Production
31mWhat does "high priority" mean? The secret to happy queues
31mA JVM threading model for the containerized times