Triggering replication

Question

How does the coherence framework decide when replication should be triggered on a cached object ? Do we have to invoke put() to ensure that any changes to the object (if any) are replicated to peer caches ?
     The objects we wish to cache have large object graphs, of which any of the underlying elements may change based upon events in our system, or based upon user interaction. The object is heavy on reads and heavy on writes.
     Thanks.

Answer 1

How does the coherence framework decide when     > replication should be triggered on a cached object ?
     > Do we have to invoke put() to ensure that any
     > changes to the object (if any) are replicated to
     > peer caches ?
     When using the Coherence API, yes.
     When using Coherence*Web session management, no.
     > The objects we wish to cache have large object
     > graphs, of which any of the underlying elements may
     > change based upon events in our system, or based upon
     > user interaction. The object is heavy on reads and
     > heavy on writes.
     Objects with large object graphs are not optimal for Coherence to deal with.
     You can use entry processors (see the InvocableMap API) to leave them "in place" and access/manipulate them remotely. That is optimal for heavy writes, and (due to the number of changes) acceptable for heavy reads.
     Breaking up the graph, if conceptually possible, would be a more optimal approach. Are you already working with an account manager at Tangosol, since they can allocate appropriate engineering resources to help you with this? Would you want to set up a discussion with one of our architects to determine an approach?
     Peace,
     Cameron Purdy
     Tangosol Coherence: The Java Data Grid