NCS-Core-6.8 Sample Questions & Answers

To support tagged VLANs for CVM and hypervisor traffic, the physical switch ports must be configured as trunk ports (or the equivalent terminology for the switch vendor). A best practice is to set the native (untagged) VLAN to a dedicated, unused VLAN ID. This ensures that no untagged traffic inadvertently ends up on a production VLAN and forces all traffic, including management, to be explicitly tagged. The management VLAN ID will then be specified during the Foundation process or node configuration.

The NCC output provides detailed information for each check, including the reason for failure and the thresholds used. Before taking any corrective action, the first and most logical step is to analyze this output. It will clarify if memory usage is slightly above a warning threshold or critically high, and it often provides links to relevant knowledge base articles. This information is crucial for determining the correct next action, which might range from adjusting CVM memory (if supported for the enabled features) to identifying a specific memory-consuming process.

Firmware upgrades for certain hardware components, like SATADOMs, cannot be performed while the main AOS is running. The automated LCM process handles this by rebooting the CVM into a special maintenance environment called Phoenix. From Phoenix, it can safely apply the low-level firmware update. The LCM process orchestrates this one node at a time to maintain cluster availability.

The balance-tcp mode in AHV networking uses LACP (Link Aggregation Control Protocol) and load balances traffic based on a hash of Layer 4 information (source/destination IP and TCP/UDP port). This provides a more granular and effective distribution of traffic from multiple VMs and sessions across all active uplinks in the bond, making it the ideal choice for maximizing throughput in a high-traffic environment. balance-slb balances based on source MAC address, which can lead to uneven link utilization. active-backup only provides redundancy, not load balancing.

This solution correctly addresses all constraints. An Asynchronous Protection Domain with a 60-minute schedule meets the 1-hour RPO. In-transit encryption can be enabled on the remote site configuration to satisfy the security requirement. Critically, a remote site throttling policy allows the consultant to limit the bandwidth used by replication during business hours (9-5) to avoid impacting other critical traffic, while allowing it to use more bandwidth overnight. Metro Availability is not feasible due to the 25ms latency. NearSync is unnecessary as the RPO is 60 minutes. Cloud Connect requires a cloud subscription, which was not mentioned.

In AHV networking, a managed network (VLAN) that is not associated with an uplink bond creates an internal-only, or host-only, network. VMs connected to this network can communicate with other VMs on the same host within that same network segment, but because there is no physical uplink, they are completely isolated from the external network and other VLANs. This is the simplest and most direct way to achieve the required isolation for a security sandbox.

Zookeeper is a critical service in the Nutanix architecture that provides coordination and acts as a distributed, consistent configuration store for all cluster services. It ensures that all nodes have a consistent view of the cluster's state, manages master-slave roles for other services, and handles leader election. It does not manage the data I/O path (that's Stargate) or store VM data (that's Cassandra for metadata and the extent store for data).

The As-Built document should map the logical Nutanix configuration to the physical environment. The physical switch port connections are part of the customer's infrastructure and are not discoverable by the Nutanix software. This information must be recorded manually.

Similar to switch port information, the physical location of the hardware (rack number, U-position) is external to the Nutanix cluster's configuration data and must be documented by the consultant to provide a complete picture of the deployment.

In modern versions of AOS, the cache mode setting is a legacy artifact from very early versions and is no longer used. The Stargate process now dynamically determines the optimal I/O path and caching behavior for all workloads automatically. Therefore, seeing this setting as bypass has no bearing on performance, and the administrator's conclusion is incorrect. The root cause of the latency lies elsewhere.

High availability for Prism Central is achieved by deploying a scale-out instance, also known as PC HA. This consists of a three-VM deployment. These three VMs form a resilient cluster that can tolerate the failure of one VM without losing management capability. This is the standard, built-in method for making Prism Central highly available. It can be deployed on a single Nutanix cluster.