When planning your replication system, take these considerations into account:
When one of the remote sites is unavailable because of a network failure, it is likely that the other sites on the same network will be unavailable as well, so many queues will grow simultaneously. Thus, you need to allocate enough disk space for all queues to withstand a simultaneous failure of the planned-for duration.
When the queues start filling up, you can always add more partitions if you have spare disk space.
If all the queues fill up and the inbound queue cannot accept more messages, the primary database will not be able to truncate its log. If you manually override this restriction, replicate databases will not receive the truncated updates.
