The Next Frontier in Open Source Java Compilers: Just-In-Time Compilation as a Service
summary
OpenJ9’s JITServer (Samaru Cloud Compiler) separates JIT from the JVM, runs as an OpenJ9 persona in containers and Kubernetes, and demonstrably cuts node memory by 6.3 GB (≈33% fewer nodes) while keeping steady state throughput equal to baseline. Demo shows a JITServer client hitting ≈2,500 ops/s versus a memory‑constrained JVM at ~1,000 ops/s. Recommended config: 1–2 GB server memory, 10–20 clients per server and ~1 ms network latency for best ramp‑up and autoscaling behavior.
AI-written summary · about
chapters (8)
- 0:00JIT as a Service intro
- 4:16Traditional JIT challenges
- 9:58JIT Server technology
- 13:32Performance benefits
- 17:09Live demo setup
- 22:09Using JIT Server
- 27:29Recommendations and conclusion
- 31:02Q&A
related talks
29mThe Next Frontier in Open Source Java Compilers: Just-in-Time Compilation as a Service
40mNext Frontier in Open Source Compilers: Just-In-Time Compilation as a Service
33mCloud Native Java with OpenJ9: Fast, lean and definitely mean
55mJIT and AOT in the JVM
52mCloud Native JVM. Cloud Compiler
41mThe Cloud Native Compiler: JIT-as-a-Service
14mCloudJIT: A Just-in-Time FaaS Optimizer (Work in Progress)
49mJava and Containers Make it Awesome