You are standardizing storage policies in Prism Central for three AHV tiers: OLTP database VMs need predictable low latency, general application VMs generate bursty I/O that must not starve other tiers, and file-share VMs prioritize capacity. Nutanix storage QoS sets throttle limits only. Which storage-policy mapping best reflects these requirements?

Prism Central storage policies are associated with VMs through categories and can set encryption, compression, QoS and replication factor. Nutanix QoS is a throttle: it caps IOPS or throughput for the VMs in the policy, and there are no priority tiers or minimum guarantees. For the OLTP tier the right move is therefore to leave it unthrottled so it can use the performance it needs. The bursty application tier is the potential noisy neighbor, so a moderate QoS limit keeps its spikes from crowding the database out. File shares care about capacity, so a policy with compression enabled saves space where latency matters least. Putting compression and dedupe on the database while leaving the application tier unthrottled protects nothing, and RF3 on file shares adds a third copy of data, which costs capacity rather than saving it. Throttling the application tier is right, but giving the database compression instead of protection and leaving file shares uncompressed misses two of the three goals. Throttling the database caps the workload you're trying to protect, which inverts the design. Exam caveat: QoS on Nutanix means a ceiling, never a floor, so you protect important VMs by limiting the others. Operational check: confirm category assignments, then compare database latency and the application tier's capped IOPS in Prism during peak.