For large cluster sizes, non-request serving pods such as OVN, etcd, etc. require more resources. Because these pods live in the same nodes as other hosted cluster's non-request serving pods, we can run into resource exhaustion unless requests for these pods are properly sized.  

    4.16

    Always

    1. Create large hosted cluster (200+ nodes)
    2. Observe resource usage of non-request-serving pods
    

    Resource usage is large, while resource requests remain the same

    Resource request grows corresponding to cluster size