Thawing Java on AWS Lambda
summary
A simple Spring Cloud Function on AWS Lambda starts with ~11–16 second cold starts, then moving the code to Micronaut reduces cold starts to about 8.5s, and compiling with GraalVM Native Image yields ~800ms cold starts and much lower memory use. The demo shows replacing Spring Cloud Function, swapping annotations, measuring 100 forced cold starts, and hitting native-image pitfalls (long compile time, reflection config, platform-specific binaries) but achieving subsecond cold startup.
AI-written summary · about
chapters (9)
- 0:00Start
- 7:27Cold Start Call Graph - Theory
- 8:30Benchmarking approach
- 11:29Spring Startup
- 13:51Compile Time vs Runtime
- 15:15Very similar to work with
- 17:32Rust looks very different from Java
- 18:18Cold Start Benchmarks
- 22:21JVM vs Native
related talks
46mThawing Java on AWS Lambda Reducing Cold Start Times from 11 Seconds to 1
46mReduce Cold Starts for Java Serverless Applications in AWS
51mHow To Reduce Cold Starts for Java Serverless Applications in AWS
44mHow to develop, run and optimize Spring Boot 3 application on AWS Lambda
19mTen things that improve the performance of your Java based AWS Lambda
56mAfraid of Java cold starts in Serverless? Fear not, Java is fast
53mHigh performance Serverless Java on AWS
49mServerless Java Web Applications on AWS Lambda with Micronaut Quarkus & Spring Boot