We have large dataset around 40Million+ records in the database (and growing). The deployment is done in multiple datacenter and all nodes are connected. Each datacenter with 3 node cluster.
The table consists of composite PK with a string (64 chars) and an integer field. The operations that happen on this table are insert and reads (update and delete are really rare). Compared to inserts, select are more in number but read time is so huge. Always SELECT query uses PK to fetch the record.
Please see below screen shots. Initially this was good and we didn’t see any problem. Able to get around 1100 reads/sec but it hardly 100 that are successfully fetched. Not sure where the issue is.
Let me know if you guys have any recommendation about this or possible cause. Version used is 19.1.4