Hey,
Well, there currently is no plans on our side to support replication as you know it from the 2.x line.
Now, we'll have clustering support backed by Terracotta's new platform that's currently being worked on. Clustering comes with many advantages over the replication models: proper consistency model(s), better data locality on a per-node basis, ... All OSS and free to use.
While, again, we don't plan on actively working on such an approach ourselves on the short term for the 3.x line, maybe others may show interest. The modular approach of the 3.x line should make this not only perfectly feasible, but also seamless for users to then use.
Alex