Twitter's quest for a wholly Graal runtime

on YouTube ↗50m

" summary

Twitter integrates Graal into a custom JDK 8 (via JVMCI) and replaces C2 on key services (Finagle/Scala-heavy tweet, user, timeline) to get concrete wins: fixes for a lock-related bailout, an agent-instrumentation SIGSEGV, and a reverseBytes intrinsic bug; ported compact-string intrinsics into Graal; ran the tweet service 100% in production for >1 year with zero crashes; measured 11% CPU reduction (≈2.5% fewer GC cycles), another ~2% from inlining tuning, and estimated large annual savings at fleet scale.

AI-written summary · about

" chapters (15)

1. 0:00Introduction
2. 5:58Twitter's infrastructure and JVM challenges
3. 8:10Graal over C2 choice
4. 13:09Graal bugs in production
5. 25:37Graal stability and adoption
6. 27:02Contributing to OpenJDK
7. 29:51Graal vs C2 benchmarks
8. 37:14Scaling Graal at Twitter
9. 40:03Cost savings from Graal
10. 42:55Fine-tuning Graal
11. 44:44Future of Graal and JVM
12. 45:31Graal in JDK 10+
13. 46:12Graal learning resources
14. 47:17Project Metropolis and JVM
15. 49:32Summary and call to action

" related talks

• 41mTwitter’s Quest for a Wholly Graal RuntimeChristian Thalinger · Oracle OpenWorld 2017
• 53mTwitter's quest for a wholly Graal runtimeChris Thalinger · JavaDay Ukraine 2017
• 54mTwitter's quest for a wholly Graal runtimeChris Thalinger · Devoxx Belgium 2017
• 56mPerformance tuning Twitter services with Graal and Machine LearningChristian Thalinger · Voxxed Days Cluj-Napoca 2019
• 54mPerformance tuning Twitter services with Graal and Machine LearningChris Thalinger · Voxxed Days Milano 2019
• 40mScala ♥ GraalFlavio Brasil · QCon São Paulo 2019
• 41mGraal: How to use the new JVM JIT compiler in real lifeChris Thalinger · Scale By The Bay 2018
• 34mPerformance Tuning Twitter Services With Machine Learning, Part 2 JIT and GCChri · Devnexus 2020