A finance VM on AHV must maintain disk I/O during month-end, when batch-processing VMs on the same cluster generate heavy storage load that starves it. The administrator needs to protect the finance VM's disk performance. Which setting should be applied?

Storage QoS on Nutanix is delivered through Prism Central storage policies, which are associated with VMs through categories. The QoS setting is a throttle: it limits the IOPS or throughput the covered VMs can consume. There is no IOPS floor or reservation, so the way to protect a critical workload is to cap the workloads that crowd it out. Tagging the batch-processing VMs with a category tied to a throttling storage policy keeps their month-end burst from saturating the CVMs and disks, which restores headroom and lowers latency for the finance VM. Adding vCPUs to the finance VM addresses CPU starvation, not disk contention, and may even increase its own I/O demand. Enabling HA protects against host failure by restarting VMs elsewhere; it does nothing for performance while the host is healthy. Raising the container's replication factor adds a copy of every write, increasing back-end I/O and capacity use, which makes contention worse. Exam caveat: storage policy attributes are encryption, compression, QoS throttling and replication factor; a minimum-IOPS guarantee does not exist. Operational check: after applying the policy, watch the batch VMs' IOPS hold at the limit and confirm the finance VM's disk latency improves during the next batch run.