we use karaf and hazelcast together. The hazelcast instance has a backing database map-store and is offered as a osgi service. In some deployments, for 2 groups of tables, the database map-store is the same for each group but in some cases different databases are used. Many karafs (up to 10) are deployed with this configuration and each will have hazelcast instances.
Does it make sense to optimize the deployment so that in the case that only one backing map-store exists, only one hazelcast instance is used? Or is the overhead of 2 instances of hazelcast per karaf neglegible enough that this effort is not worth the trouble?