Nutanix Metro Availability and Synchronous Replication have very strict networking requirements to ensure that real-time data mirroring does not introduce unacceptable application latency. The industry-standard requirement for Nutanix synchronous operations is a maximum Round Trip Time (RTT) latency of 5ms between the two primary clusters (Site 1 and Site 2). According to the diagram provided in the question, the latency between Site 1 and Site 2 is 25ms RTT. This latency is significantly higher than the 5ms threshold and will prevent the Metro Availability configuration from functioning correctly. While the latency to the Witness VM (200ms RTT) is perfectly acceptable (Option D), and having a Witness at a third site (Option A) is a best practice, the " data path " between the storage clusters is the bottleneck here. If an administrator attempts to enable Metro with 25ms of latency, the guest VMs would experience severe performance degradation because every write would be delayed by the 25ms it takes to receive an acknowledgment from the remote site.
Therefore, the high latency between the data-serving sites is the primary configuration issue that must be resolved, typically by moving the clusters closer together or upgrading the inter-site fiber connection.

In a Nutanix legacy Protection Domain (PD) configuration, three primary elements must be in place for data protection to occur: the entities (VMs), the remote site (destination), and the schedule. Even if a VM has been added to a Protection Domain (negating Option A), the cluster will not take any action unless it is explicitly told " when " and " how often " to do so.
If no new snapshots are being created locally and replication is not starting, the most likely issue is that no " Schedule " has been configured for that Protection Domain. The schedule defines the RPO (e.g., hourly, daily) and the retention period for the snapshots. Without an active schedule, the Cerebro service simply has no instruction to initiate the snapshot process. While a missing vStore mapping (Option C) would cause the " replication " to fail, the " local snapshots " would still be created in the PD. Similarly, low bandwidth throttling (Option D) would cause the replication to move slowly or time out, but it would not prevent the initial creation of the local recovery points. Therefore, verifying the presence of a valid, active replication schedule is the mandatory first step for an administrator when a Protection Domain appears to be " idle " despite having VMs assigned to it.

Nutanix Metro Availability is a continuous availability solution designed to survive a site failure with zero data loss (Zero RPO). Because it operates on a synchronous replication model, every write performed by a virtual machine must be acknowledged by both the local and the remote clusters before the write is confirmed to the application. This process introduces a " latency penalty " equal to the network Round Trip Time (RTT) between the sites.
To maintain acceptable application performance and ensure cluster stability, Nutanix enforces a strict latency limit of 5 ms RTT between the primary and secondary clusters. If the latency exceeds this 5 ms threshold, the time spent waiting for remote acknowledgments would cause severe performance degradation for high-I/O workloads, such as the SQL or Oracle databases common in financial institutions. Furthermore, high latency increases the risk of the synchronous link being " broken " during periods of high data churn. While the Witness VM (used for tie-breaking) can tolerate much higher latency (up to 200 ms), the data-path link must remain under 5 ms to satisfy the technical requirements of the Metro Availability feature and the performance expectations of Tier-1 business-critical applications.

When a replication job is marked as " ongoing " but the progress (bytes completed) is stagnant, it typically indicates a mismatch in the configuration or a failure in the orchestration logic rather than a total network blackout, especially when virtual IPs are confirmed reachable. In a " new setup, " a very common root cause is an incorrectly configured Protection Policy.
This can include issues such as mismatched category assignments, incorrect target site selection, or an invalid replication schedule that the system is unable to initialize properly. If the policy itself is misconfigured-for instance, if it is attempting to replicate to a storage container that hasn ' t been mapped or lacks permissions- the Cerebro service may initiate the task but then fail to progress the data transfer. While blocked ports (Options C and D) are common culprits in replication failures, they usually result in immediate " unreachable
" alerts or job failures rather than a " stuck " status where connectivity is otherwise reported as healthy. If the Acropolis service were down at the remote site (Option A), the cluster would likely report the site as disconnected or show a critical service alert in Prism. Therefore, the most logical first area to investigate in a new, non-progressing setup is the logic and settings defined within the Protection Policy itself.

Nutanix manages data recovery operations differently depending on whether a virtual machine is in an asynchronous or synchronous replication state. For a standard Asynchronous VM like " REPORTING-DEV, " the local cluster takes periodic snapshots. If data corruption occurs, the administrator can perform an " In- place Restore " or " Revert " using a local recovery point. This is a standard procedure where the current vDisks are overwritten by the point-in-time data from the snapshot.
However, " ERP-PROD " is part of a Synchronous replication policy (0 RPO). This means the storage container it resides on is effectively " stretched " and locked in an active-active relationship with a remote cluster. Every write must be identical at both sites. If an administrator attempts to " Revert " a synchronous VM to a local snapshot, it would cause an immediate data mismatch between the primary and recovery sites, breaking the zero-RPO guarantee and potentially causing cluster-wide consistency issues. To protect data integrity, Prism Central typically grays out or blocks the " Revert " option for VMs currently in an active synchronous " Sync " state. To recover such a VM, an administrator would generally need to either " Clone " the snapshot to a new VM or temporarily move the VM into an asynchronous state by modifying the policy, thereby unlocking the ability to revert the data disks locally before re-enabling synchronization.