I love the small memory of the library I embedded (since I have many issues with the memory of my pc). It is also very fast and not complex at all and it saved my all my problems in the distributed systems for sure!

Performance and how it is easy to start implementation. In-memory processing is great adventage, because of high availability.

Hazelcast first of all is a really great distributed data grid. You can use distributed collections out of the box like Map, Set, Queue for microservices replicas to share data or for caching purposes. Also, it is possible to take distributed locks to prevent doing the same work simultaneously. The Streaming framework Hazelcast Jet has many needful features and is very fast. Can be run separately from services or as a library. Supports automatic rebalancing and replication. Multiple network options to make a cluster such as multicasting or kubernetes support.