Description of problem: While running scale tests of OpenShift on OpenStack at scale, we're seeing it performing significantly worse than on AWS platform for the same number of nodes. More specifically, we're seeing high traffic to API server, and high load for the haproxy pod.

Version-Release number of selected component (if applicable):

All supported versions

How reproducible:

Steps to Reproduce:

1.
2.
3.

Actual results:

Expected results:

Additional info:

Slack thread at https://coreos.slack.com/archives/CBZHF4DHC/p1669910986729359 provides more info.