Performance tuning Twitter services with Graal and Machine Learning
summary
Running Twitter services on Graal combined with a Bayesian-optimization service called Autotune targets Graal inlining parameters and shows concrete wins: about 30% user‑CPU reduction vs C2, with Autotune adding roughly 6% more (total ~18% in some runs), GC cycles cut and end‑to‑end latency improved up to ~28% after tuning. The system is a driver that runs experiments, measures steady‑state 30‑minute evaluations on dedicated machines, and rejects configurations that violate throttling constraints.
AI-written summary · about
chapters (12)
- 0:00Start
- 2:25ABOUT ME...
- 6:13PERFORMANCE OPTIMIZATION
- 7:36PERFORMANCE TUNING
- 8:45OPTIMIZATION OF A BLACK BOX FUNCTION
- 12:49WHAT IS AUTOTUNE?
- 14:01WHAT IS GRAAL?
- 15:41WHICH PARAMETERS?
- 19:32WHAT RANGES?
- 20:20TEST SETUP
- 22:53TWEET SERVICE: CONSTRAINTS
- 31:09SUMMARY
related talks
54mPerformance tuning Twitter services with Graal and Machine Learning
49mPerformance tuning Twitter services with Graal and Machine Learning
46mPerformance-Tuning Twitter Services with Graal and Machine Learning
56mPerformance tuning Twitter services with Graal and Machine Learning
01h 01mContinuous Optimization of MIcroservices using Machine Learning
29mContinuous tuning of Microservices via Bayesian Optimization
22mAutomating Performance Tuning with Machine Learning
18mLet the machines optimize the machines