Caching beyond simple put and gets
summary
Simple tapping leaderboard app hits write-scaling limits once users grow; dual writes (app cache plus SQL) create stale, unscalable state. The talk shows concrete fixes: move from cache-aside to read-through/write-through, implement persistence via Hazelcast MobStore or a JDBC mapping so cache persists synchronously or asynchronously, use Viridian for managed clusters, and rely on built-in replication, TTL eviction and entry processors. A live demo fails but the implemented patterns and trade-offs (consistency, latency, backups) are shown and evaluated.
AI-written summary · about
chapters (21)
- 0:00Introduction and setup
- 1:09Why caching matters
- 2:58James's tapping startup story
- 7:43Cache-aside pattern problems
- 10:33Scaling embedded caching issues
- 11:35Separate cache instance drawbacks
- 12:38Minimizing inconsistency gaps
- 13:21Read-write through pattern
- 15:46Cloud-based caching solutions
- 16:55Separation of concerns benefits
- 18:20Offloading to caching platforms
- 19:03Hazelcast MapStore demo explanation
- 21:11Third-party code confidence
- 21:53Read-through cache pattern
- 22:33Legacy application caching
- 23:35SQL as universal data language
- 26:02Minimal app changes approach
- 26:24Caching patterns recap
- 27:05Advanced in-memory computations
- 29:35In-memory vs database convergence
- 29:56Q&A session begins
related talks
43mCaching and Beyond - Smarter Data Processing with Java
30mCaching patterns for lazy devs for lazy loading
46mCaching Made Bootiful
46mA Hitchhiker’s Guide to Caching Patterns
45mCaching in Applications Still Matters
43mWhere Is My Cache? Architectural Patterns for Caching ...
42m3 easy performance improvements in your microservices architecture
46mEdge Performance with In memory NoSQL