You need to test a Windows patch on a production AHV VM without changing the running source workload. Which action gives you a disposable copy for the test?

On AHV, cloning a VM or a snapshot of it creates a new, independent VM whose disks start from the same data. The cluster uses copy-on-write metadata, so the clone is created almost instantly and only consumes space for blocks it changes. That gives you an isolated test VM: patch it, test the application, and delete it, while the production VM keeps running with its disks untouched. Cloning from a snapshot also lets you pick a known point in time. Taking a snapshot, patching production and reverting is not a copy strategy; it changes the live workload, and reverting discards any production writes made since the snapshot. Converting the production VM into a template and redeploying it from a patched template replaces the source VM with a modified version, which violates the requirement not to change it. Exporting an OVA, patching the copy and re-importing it over the source is a destructive replacement path with downtime. Exam caveat: AHV doesn't use vSphere's linked-clone terminology; Nutanix clones are always space-efficient because of how AOS stores data. Also give the clone a different network or disconnect its NIC so a duplicate hostname or IP doesn't conflict with production. Operational check: verify the clone powered on as a separate VM, patch only the clone, and delete it when the test is complete.