Describing how going off-heap with the help of Chronicle Map helped us to get rid of the latency hiccups and cut our Amazon AWS bill in half.
Caution: Before you go diving into using the open source version of Chronicle Map, there are some gotchas to be aware of from a features point-of-view.
Chronicle Map could be persisted to disk. **Durability with Chronicle Map is provided by another level of architecture,** for example all requests are sent to several nodes - master and hot standby. Clustering/distributed architecture is out of the scope of the Chronicle Map project, there are projects _on top_ of Chronicle Map which address these questions, e. g. [Chronicle Enterprise](http://chronicle.software/products/chronicle-enterprise/).
If your needs are to persist off-heap data on the local JVM to disk (and to have multiple processes interact with data within one physical machine) then you can use the open-source edition but if your needs are to talk across JVMs that are across IPs (physical/virtual machines) for synchronization/replication features then you are looking at the closed source (paid) version of their product stack as seen from this link. Your needs may vary.