Clustering : Clustered Data Store: deleting of large number of flows locks up the controller

After inserting 50k flows into config data store (no mininet attached), i attempt to delete the flows in batches of 5k flows, where 10 concurrent threads execute the delete operations over Restconf. Delete start with acceptable rate (360 flows/sec), but trickles down to 2 flows/sec in batches 8 and 9, and the controller locks up during batch 10. This problem does not occur with the in-memory data store.