Arc-enabled SQL MI and Kubernetes work together such that a standard deployment will distribute the instances across the worker nodes in the cluster, and that the replicas for a given instance will run on different workers where possible. However, to demonstrate the power of a single worker node to run all pods in the workload, we took the other nodes offline, so Kubernetes was forced to run all of the workload on our PowerEdge R650xs server. This is not typical and was done for illustration purposes of power and scale of a single server.
