Also, how will you rebalance a live cluster, given Fabric does not have LB replication like a multi…
R Krishna

As stated in my previous response, all Kafka does is temporarily store the blocks of transactions, orders them and sends them to the OSNs to persist. So on rebalance, unless you have transactions that are yet to be ordered and sent to the OSNs for delivery, rebalance should be fine. Therefore, there must be downtime, unless Fabric comes up with something to counter this. I may be wrong, as this is just a guess.