Lean Spring Boot Applications for the Cloud by Patrick Baumgartner

Upgrading to newer Java versions (like Java 21) generally improves startup time and performance, with each minor version typically bringing optimizations

Using buildpacks instead of Dockerfile is recommended as they:

• Create more secure containers
• Handle dependencies automatically
• Are maintained by the community
• Make container creation easier

Spring Boot optimizations:

• Remove unused dependencies and starters
• Make Spring Beans lazy where possible
• Disable unnecessary actuators
• Use AOT (Ahead of Time) processing
• Configure proper memory settings

Class Data Sharing (CDS) can improve startup time by:

• Dumping loaded classes into a file
• Loading classes from the dump file on subsequent startups
• Reducing class loading overhead

Virtual threads (Java 21+) provide benefits:

• Better performance for web applications
• Improved response times
• More efficient resource usage
• Work well with 1-4 CPU cores

JVM runtime options that help:

• Using tiered compilation with stop level 1 for faster startup
• Proper garbage collector selection
• JLink to create custom minimal JVMs
• Configuring proper heap and memory settings

Container image size can be reduced by:

• Using minimal base images (like tiny)
• Removing unnecessary modules
• Using JLink to create custom JVMs
• Cleaning up dependencies

Benchmarking best practices:

• Do warm-up runs
• Test with different CPU/memory configurations
• Measure startup time, throughput and memory usage
• Consider your specific use case
• Don’t trust generic benchmarks

OpenJ9 can be a good alternative JVM when memory optimization is important

Application packaging recommendations:

• Deploy exploded (unzipped) applications for faster startup
• Use proper configuration locations
• Consider environment-specific optimizations