Kotlin on CRaC – new JVM features to speed things up

on YouTube ↗40m

" summary

Shows concrete numbers and how to reproduce them: CRaC (Coordinated Restore at Checkpoint) with an Azul Zulu CRaC JDK on Linux plus Docker cuts a Spring Boot Kotlin app from ~2.7s cold start to ~0.5s restore (benchmarks sometimes hit ~100ms) using Siege-driven warmup and a saved checkpoint; JEP 483 AOT class loading and JEP 515 AOT method profiling produce an AOT cache that drops a Spring run from ~2.1s to ~1.15s; mocking demo goes ~500ms → ~40ms; limitations: Linux-only CRaC, CRaC requires close/open hooks, AOT needs stable jar-only classpaths so Gradle workflows need workarounds.

AI-written summary · about

" chapters (10)

1. 0:00JVM performance challenges
2. 3:12Slow JVM startup reasons
3. 8:50What is CRaC and how it works
4. 12:00Live CRaC demo with Spring Boot
5. 21:27CRaC backend benefits
6. 22:32CRaC for unit tests
7. 26:47AOT cache benefits
8. 31:49CRaC with Gradle tests
9. 37:47Future JVM performance
10. 39:12Summary

" related talks

• 53mJava on CRaC: Superfast JVM Application StartupSimon Ritter · Jfokus 2022
• 51mJava on CRaC Superfast JVM Application StartupGerrit Grunwald, Simon Ritter · Devnexus 2022
• 46mCRAC vs GraalVMLilian Benoit · Devoxx France 2023
• 45mWhat the CRaC - Superfast JVM StartupGerrit Grunwald · DevDays Europe 2024
• 53mWhat the CRaC Superfast JVM startupGerrit Grunwald · Voxxed Days Cluj-Napoca 2022
• 45mOptimizing Java Startup and Warmup: Exploring the Power of CRaC on EmbeddedFrank Delporte · JCON WORLD ONLINE 2023
• 52mDuke on CRaC with Jakarta EEIvar Grimstad, Rustam Mehmandarov · JCON EUROPE 2025
• 53mFour Approaches to Reducing Java Startup TimePasha Finkelshteyn · Confitura 2025