I have a shared-data cluster running 3.4; when I run

SHOW DATA

I'm seeing ReplicaCount on the order of 9396 for a variety of tables -- this goes against my intuition that the replica count should be much closer to 1 or at least match the

"replication_num" = "1",

defined on the table? What am I missing, or what should I look at for further debugging?

replicaCount for a table = number of tablets of the table x replication_num per tablet.