MapDB - taking Java collections to the next level
summary
MapDB is an embeddable Java storage engine that exposes maps and sets with pluggable backends, on-heap or off-heap serialization, and index/compression strategies. It ships as a 600KB jar, trades raw speed for features at about 10x slower than HashMap but adds self-expanding hash maps, expirations, modification listeners, and segment concurrency. BTree maps use primitive specialization and delta/prefix compression to fit ~20x more entries/GB and a streaming bulk loader achieves up to 39 million inserts/sec. MapDB2 is an active rewrite with further optimizations.
AI-written summary · about
chapters (8)
- 0:00Introduction to MapDB
- 1:53MapDB vs Java collections
- 5:30Expiration and caching features
- 9:30TreeMap and memory efficiency
- 13:24Data pump and bulk inserts
- 17:37Collection listeners and secondary maps
- 19:51MapDB 2.0 and future plans
- 23:01Conclusion and Q&A
related talks
48mHidden Pearls for Java Embedded High-Performance Persistence
57mJava In-Memory Database-Apps with EclipseStore
01h 12mHidden pearls for high-performance-persistence in Java
43mCaching and Beyond - Smarter Data Processing with Java
44mJava In-Memory Database-Apps with MicroStream
57mDo It Yourself: Collections
53mJava In-Memory Database-Apps with EclipseStore
32mJava's Ignored Potential: Supercharge Database Apps with Java In-Memory Process