In Prism Central VM efficiency and behavioral-learning views, Nutanix groups inefficient VMs into categories such as over-provisioned, inactive, bully, and constrained. A constrained VM is one that does not have enough resources for its workload and is therefore experiencing pressure. Nutanix documentation ties constrained status to signs such as CPU readiness, memory swapping, or other stress indicators that show the VM is not adequately provisioned for what it is trying to do. That means the best interpretation is that those VMs are lacking sufficient resources, which is option B. ( Nutanix Portal ) This category is different from the other distractors. An inactive VM is mostly idle and not making meaningful use of its assigned resources. An over-provisioned VM is wasting capacity because it was allocated too much. A VM that is merely "consuming too much CPU" could be a bully VM or simply a busy VM, but that alone does not define the constrained classification. Nutanix uses the constrained label to signal a right-sizing opportunity in the direction of adding or adjusting resources, not reclaiming them. Therefore the most accurate answer is B, because constrained VMs are the ones that are short on sufficient resources for their workload behavior. ( Nutanix Portal )

Nutanix documents guest customization profiles as reusable operating-system customization definitions that can be applied during VM deployment. This is precisely what makes them the most efficient answer when a Windows VM will be cloned repeatedly and each clone requires OS-level customization such as naming, identity settings, or first-boot configuration. Instead of creating a separate template each time or manually editing each deployed clone, the administrator creates one reusable customization profile and applies it across deployments. That makes C the correct answer.
This is especially important at scale. Manual post-clone configuration is error-prone and time-consuming, while per-deployment templates create management sprawl. Nutanix's design encourages reusable policy- or profile-based standardization so that repeated deployments remain consistent and secure. For Windows, that typically means Sysprep-oriented OS customization combined with reusable deployment logic. In other words, the question is asking for the most efficient method, and Nutanix's documented answer to repeatable OS customization is the guest customization profile. Therefore C is the authentic choice.

Nutanix documentation for PD-based DR remote-site configuration states that if vStore mapping is not configured, replication occurs to the remote container with the same name as the source container. Nutanix further explains that this fallback relies on the existence of that matching destination container. Therefore, if same-name containers do not exist on the two sites and no explicit vStore mapping is defined, the replication target cannot be resolved correctly and the replication will fail. That makes A the correct answer.
This behavior is intentional. Nutanix does not randomly pick a destination container, nor does it silently force all replications into the default container. When administrators want different container names between source and target, the supported mechanism is vStore mapping. Skipping that mapping only works when identical container names already exist at both sites. Because the question explicitly removes that condition, replications fail to complete, which is why A is the authentic answer.

The uploaded source file shows the answer key as B, but the supporting exhibit or calculation context is missing from the provided document, so a full mathematical derivation cannot be verified from the question bank itself. Because of that missing context, the safest exam-faithful response is to preserve the source answer of 1125 IOPS while being transparent that the exact intermediate values used to reach that number are not visible in the provided material.
From a Nutanix policy perspective, IOPS throttling is implemented through storage policies to control or cap workload consumption and prevent noisy-neighbor impact. The conceptual goal is to set a value high enough to accommodate the intended steady-state workload while still enforcing policy boundaries. In practice, administrators usually choose a value above the observed average to allow normal variance, bursts, and scheduler overhead rather than matching the average exactly. That design logic is consistent with an answer larger than 800 IOPS. Given the choices and the source key, 1125 is the preserved answer for this batch, but this is one of the questions where the missing exhibit should ideally be restored before using it as a final study reference. ( Nutanix )

The key clue is the mismatch between the VM being slow under memory overcommit and the guest OS pagefile usage remaining low. If the guest itself were under strong memory pressure, you would expect higher pagefile or swap-file activity inside the OS. Instead, the symptoms suggest memory reclamation is happening outside the guest, which points to hypervisor swapping. In Nutanix AHV terms, that means the platform is reclaiming memory at the virtualization layer, causing performance degradation even though the guest does not show heavy internal paging. Therefore C is the most likely cause.
The distractors are weaker. "CVM swapping" would indicate pressure in the storage/control path VM, not necessarily this guest's memory experience. Ballooning and swapping are not mutually exclusive, and the question's evidence does not specifically prove ballooning is disabled. This is a troubleshooting pattern- recognition question: slow VM, full memory use, low guest pagefile use, and overcommit enabled usually indicates the hypervisor is swapping, so C is the best answer.