Sharing Hazelcast Cluster across multiple development team

Question

Read from a few places which indicate mapstore and maploader must be running with the Hazelcast Node. Would like to find out if there any ways to implement mapstore/maploader separate from the Hazelcast Node?

Background: If i have a hazelcast cluster for the team, and this cluster is to be use by different sub-team providing different map as data, and each sub-team should implement mapstore/maploader for the map they own, how can this be done? (Note that each sub-team have their own SVN repository)

Thanks in advance~

Answer 1

MapLoader's load() operation is only invoked on the node that would have the key when the key is missing, so there is no way to push this processing elsewhere.

However, each map can have a different MapStore/MapLoader implementation, so having a different team provide each is certainly feasible.

Exactly how you achieve this comes down to your build and deploy practices. For example, each team's classes could be in a separate jar file on the classpath. Or, there could be a single jar file constructed containing the classes provided by each team. Many ways exist!