Cache, But Cache Wisely!
summary
Demonstrates why naive session caching with Infinispan can slow things down: a client stored up to 100 MB sessions, causing serialization on every read/write, eight round trips per request and added latency often in the 160–400 ms range. Uses Prometheus and Jaeger for APM and tracing during a 100-client, ~500 req/s load test to find log statements and DB queries as real bottlenecks; adding JPA second‑level cache increased outliers to 673 ms, so trim session data and use targeted DB or JCache strategies.
AI-written summary · about
chapters (13)
- 0:00Introduction and speaker background
- 2:09Client case study and caching pitfalls
- 7:08Choosing the right caching solution
- 9:35Observability and monitoring basics
- 11:46Demo setup and naive caching approach
- 16:44Load testing the naive implementation
- 22:08Limitations of APM and need for tracing
- 27:33Adding response caching and results
- 31:28Second-level database caching attempt
- 37:27Why caching didn't solve the problem
- 38:10Modern caching alternatives
- 39:13Key takeaways and best practices
- 40:38Q&A session
related talks
45mCaching in Applications Still Matters
44mHow to speed up your application using JCache
50mCaching with Spring: Advanced Topics and Best Practices
46mCaching Made Bootiful
43mWhere Is My Cache? Architectural Patterns for Caching ...
50mLet's talk about the cache!
53mLeveraging Caching For Cloud Native Applications
36mMake It Fast