Goal

Currently, we are using AWS infrastructure with metal instances as the underlying infrastructure we run hypershift/kubevirt guest clusters on top of. This infrastructure provider limits our ability to exercise some network configurations accurately though. Specifically, we can't live migrate VMs without incurring some disruption to the workloads within the live migrated worker node when the VMs are hosted on AWS. We also lack the ability to accurately simulate certain secondary network scenarios in the AWS environment.

By transitioning to baremetads (baremetal on equinix using devscripts) we have tighter control over the environment and more flexibility than AWS provides us.