The requirement is not to keep the mission-critical VMs together or apart from each other; it is to prevent them from running on a specific set of older hosts. In Nutanix Prism Central policy terms, that is a VM-Host Anti-Affinity Policy. Anti-affinity at the VM-host level allows administrators to mark selected hosts as unsuitable targets for selected VMs, which is exactly what the scenario requires. Therefore C is the correct answer.

The distractors all target the wrong relationship. VM-VM affinity and anti-affinity control how VMs are placed relative to other VMs, not whether they can run on particular hosts. VM-Host Affinity would express a positive preference or requirement to run on a chosen host group, but the question is framed negatively: the VMs must never run on the older hosts. Nutanix’s policy model supports this exact exclusion use case through VM-Host Anti-Affinity. In a long refresh delay scenario, this is the safest operational control to keep sensitive workloads off weaker or undesirable hardware without removing those hosts from the cluster immediately. Therefore C is the authentic answer.