Details
-
Improvement
-
Status: Resolved
-
Major
-
Resolution: Fixed
-
Private Beta
-
None
-
None
Description
We currently default to 10 RPC worker threads on the TS. For read-heavy workloads, we should have at least as many workers as cores, and possibly more (so that heavy reads don't cause all of the other RPCs to queue). Or we should consider pushing long reads off of the RPC pool onto a separate pool.