Cloud Native JVM. Cloud Runtime
summary
Claim: moving JVM runtime pieces (JIT, class loading and safe early initialization, profiling) into a shared service lets many JVMs reuse compiled variants and profiles so new containers hit peak throughput immediately instead of minutes later. Prototype pieces include a Cloud Compiler that stores per-method variants and a CRaC checkpoint/restore flow to snapshot warmed images; experiments show reduced client CPU during startup and typical speculative optimizations still give ~25% extra throughput while cutting warmup and outliers dramatically.
AI-written summary · about
chapters (9)
- 0:00Introduction to Cloud Native JVM
- 4:58What is Java runtime
- 9:56JVM runtime role
- 18:21JVM compilation and profiling
- 23:12Warmup problems in cloud
- 26:39Proactive JVM model
- 37:33Cloud runtime offloading
- 40:42Cloud runtime benefits
- 43:54Conclusion
related talks
52mCloud Native JVM. Cloud Compiler
47mCloud Native JVMs, parts I and II: Fleet Profiling and Optimization
01h 02mJourney to Cloud : Leave No Java Workload Behind
01h 03mKeeping Your Java Hot.
58mEnhancing Java Applications for the Cloud
46mMonolith to Serverless - Can Modern Cloud-Native Java Runtimes Do It All
41mGraalVM, CRaC, Leyden and friends - in search of TRULY cloud-native Java
47mLegal JVM Doping in Your Containers