Thinking in Serverless Java (Code)
summary
Java runs well serverless: Quarkus + AWS Lambda boots a microprofile app in ~500ms (cold start including infra ~3.5s) and serves subsequent requests in ~6ms, letting a 2 tx/s function with 2 GB RAM cost about €15/month. Deployment produces function.zip via Quarkus extension from Maven; use HTTP API Gateway or function URLs for cheap endpoints, enable SnapStart for faster AZ starts, and prefer DynamoDB single-table design or RDS proxy to avoid connection overload. No containers, use CDK/Bicep or Maven modules for IaC and cache in /tmp while warm.
AI-written summary · about
chapters (6)
- 0:00Introduction and motivation
- 10:40Serverless vs application servers
- 23:54Lambda deployment and code walkthrough
- 33:56Performance and cost analysis
- 43:12Database and infrastructure insights
- 53:11Summary and serverless philosophy
related talks
01h 03mKeynote: Why The Future of Java in the Cloud is Serverless
52mContainerless, Bloatless, YAML-less, Serverless Java #slideless
49mJava Patterns and Practices for Serverless Applications
53mDissecting Serverless Enterprise Java Applications #slideless
53mServerless Java #slideless
51mAdapting Java for the Serverless world!
53m"Serverless" Is What J2EE Was Meant To Be
01h 00mAdapting Java for the Serverless world