Before metrics are available it is important for the cluster to be able to conclude that cluster hardware meets minimum recommendations.

fio is the perfect tool for this and would allow for init containers on bootstrap and mater nodes to perform baseline performance anaysis that can later be consumed by the operator. 

Based on an agreed threshold we could have the operator go degraded resulting in failed install. As the result of the install is stored in the kube-system/bootstrap configmap we can quickly conclude if a cluster failed bootstrap.  In which case it is not supported.