Ability to set utilization and other constraints on group or cluster level

IBM Turbonomic is currently importing admission control settings which are part on a VMware HA policy as utilization constraint affecting all host in a cluster individually. For a stretched cluster this means that a 50% maximum utilization constraint is placed for each an every host in the cluster. The desired state is applied relatively to this maximum utilization. For the default desired state setting of 70% for hosts this results a desired state for host utilization of 50% * 70% = 35%.

In fact the admission control setting is a cluster wide setting. You want to avoid to use more than given portion (e.g. 50% for a stretched cluster) of the cluster resources to ensure that all workloads can be restarted in case the foreseen number of hosts in the cluster fail or are unavailable (N+1 => 1 host, Stretched => 50% of the hosts). It does do any harm if a host is utilized more than others as long as the host isn't congested and the cluster admission control level is not exceeded. This is especially true for clusters stretched over two datacenters where a VM move cross datacenter consumes the limited datacenter interconnect network capacity.

Please introduce the cluster as an additional entity in the Turbonomic supply chain and allow to set utilization constraints on cluster level independent of host utilization constraints and the desired state on host level in order to avoid move actions that does not have any benefit of performance. As an alternative, if the cluster is not made available as entity, host groups (ideally such a group is created automatically per cluster) can be used to set utilization constraints on group level.