[DocDb] Change reactor selection on Inbound connections to round robin #879
Labels
area/docdb
YugabyteDB core features
kind/enhancement
This is an enhancement of an existing feature
priority/medium
Medium priority issue
Projects
Jira Link: DB-1536
Currently, we use a hashing scheme to select which reactor thread to schedule both incoming and outgoing connections on.
For incoming connections, this has the drawback that it could lead to 2 connections being scheduled on the exact same reactor, thus sharing 1 CPU for 2 different clients worth of load, which creates a bottleneck. Using a round robin scheme here would help fanning out across all available reactors, helping under heavy load.
For outgoing connections, we would still like to keep the current hashing scheme, as we only have a fixed set (4) of outbound connections we use and would like to ensure stickiness on those.
The text was updated successfully, but these errors were encountered: