We have group updates that are very large in size (20k+
records) and while we're specifically targeting the
engineering group responsible, there is nothing to stop
another group to write bad code and abuse our service. When
the large update occurs during replication of the group from
our master to our search fleet we often miss SLAs on
replication and search latency because the boxes go under
such heavy load.
Is there a way on our master servers we can block people from
preforming such updates? Either to discard and error on the
write or just drop that traffic all together. I'm open to any
sort of suggestion this list might have to offer.