One approach is to keep track of the ID of the last consumed event and use that as the starting point for rehydration. That way, even after a migration or failover, the stream can resume efficiently without replaying the entire history.
No comments yet.