We have recently released Infinispan 7.2, and we feel it deserves a more detailed introduction. This release brings important new features, code optimizations and major bug fixes.
What features can be found?
- We have introduced a new factory for the atomic object. The new factory is based on a replicated state machine, and it allows object creation on top of Infinispan cache e.g. Java collections.
- Server-side scripting allows custom script installation in Infinispan server and their invocation over Hot Rod. You can now use map/reduce and distributed executor over Hot Rod as well.
- JCache (JSR-107) support over Hot Rod.
- Near cache support for Hot Rod Java client.
- Support for defining filters using the Query DSL for event listeners.
- DSL-based filters are indexed for faster and more scalable evaluation.
- Custom cache stores deployment in Infinispan server.
What optimizations have been made?
- Eviction improvements with new long size support and more accurate eviction.
- Many indexing improvements.
- Faster bulk operations (putAll() and getAll()) for both embedded and Hot Rod client.
- Cache creation and configuration changes do not require Infinispan server restart.
- Faster clear operation. It is now lock-free and non-transactional, and it shouldn't be invoked with other concurrent operations.
What bugs have been fixed?
- DummyTransactionManager has been fixed and made more reliable. Although we recommend a use of a proper TransactionManager, it can be used for simpler applications which don't involve additional resource enlistments.
- Fixed deadlocks and errors during topology changes. Adding and removing nodes works smoothly and faster now.
- Multiple tweaks in CLI operations.
- Other small bug fixes.
Finally, we would like thank all community members involved in this release. A special thank you to Pierre Sutra (atomic object factory contribution), Damiano Albani (dependency fixes) and Karl von Randow (configuration fixes)
- The Infinispan team