Description of problem:
In AWS, when running node-pod-density workload to fill a worker node to maximum 250 pods per node, I am unable to reach 250 pods. Windows workers begin to be unable to stay reliably Ready above 230 pods.

Version-Release number of selected component (if applicable):
OCP 4.11
WMCO 6.0.0

How reproducible:
80%

Steps to Reproduce:
1. Provision a 4.11 cluster in AWS with 2-8 Windows workers (i.e. ovn hybridnetworking, WMCO 6.0.0)
2. Run node-pod-density workload with PODS_PER_NODE>=230

Actual results:
One or more windows workers will become NotReady.
Sometimes they will reboot themselves, other times they have remained in a NotReady state and have been unrecoverable.

Expected results:
Windows workers will remain Ready under 250 pod load.

Additional info:
Windows worker instance types tested: m5.2xlarge
load pod images tested: mcr.microsoft.com/oss/kubernetes/pause:3.6 , k8s.gcr.io/pause:3.6
node-density workload from: https://github.com/afcollins/e2e-benchmarking/commit/d09ea2accd5d9b0aed6defe6da02282840459e04
See kube-burner env values from (except set PODS_PER_NODE=240): https://github.com/afcollins/airflow-kubernetes/blob/0b7514875cb960cf0853ff9517a903cf5c1af683/dags/openshift_nightlies/config/benchmarks/control-plane-winc.json