I am porting an event-driven architecture to a container environment (Docker) and I wonder how to take advantage of replication while using pub-sub: if I just increment replicas I end up with many process doing the same elaboration and I don't think is a good way to go. I thought to replace pub-sub with producer-consumer, so consumers waits for events from a queue and only one grabs it, but I would like to avoid tightly coupling.
Is there any solution to have a "load balanced" publish-subscribe? Also commercial tools can be checked.