Designing Cisco Data Center Infrastructure (DCID) Free Sample Questions

End-Host Mode (EHM) is the correct choice. In EHM, the Fabric Interconnects behave like a large network interface card (NIC) aggregator to the upstream network. They do not run STP on their uplinks and instead rely on pinning traffic to specific uplinks, which prevents loops. Switch Mode would make the FIs act as standard Ethernet switches, participate in STP, and would not meet the design requirement of isolating the UCS domain from the upstream STP.

Multipath I/O (MPIO) is a framework that allows storage traffic to be load-balanced across multiple physical paths between the initiator (ESXi host) and the target (storage array). This is achieved at the host level, using multiple VMkernel adapters for iSCSI, and does not require special switch configurations like port-channels. It provides both redundancy and performance scaling.

The nxos_config modules (and other resource-specific modules like nxos_vlans or nxos_vrfs) are designed for declarative, state-based management. By defining the state: present or state: absent parameters, the module ensures the device's configuration matches the desired state defined in the playbook, making the operation idempotent. The nxos_command module is imperative, simply running commands without knowledge of the current state, and is less suitable for this requirement.

The BGW acts as the gateway for inter-site traffic. It decapsulates VXLAN traffic from local VTEPs.

After decapsulation, the BGW re-originates the traffic into a new tunnel across the DCI network to the BGW in the destination site. This isolates the failure domains of each site.

This statement is false. Both vNIC templates (using MAC pools) and vHBA templates (using WWNN/WWPN pools) are designed to automate the assignment of unique hardware identifiers. This is a fundamental principle of stateless computing in Cisco UCS, allowing for rapid and consistent server provisioning.

This is the comprehensive and correct design. Creating a dedicated system class for FCoE allows for specific treatment. Enabling Priority Flow Control (PFC) is essential for creating a no-drop (lossless) behavior for this class. Assigning a guaranteed bandwidth percentage ensures FCoE traffic has the resources it needs, even during congestion. The remaining traffic can fairly share the rest of the bandwidth in the default class.

The peer-gateway command allows a vPC peer switch to act as the active gateway for packets that are sent to the router MAC address of the vPC peer. This prevents traffic from being dropped when, for example, a connected device or an upstream router sends traffic to the MAC address of the other vPC peer. It enables local forwarding on the secondary peer without needing to traverse the peer-link.

N_Port Virtualization (NPV) mode allows the Fabric Interconnect to act as a pass-through device, aggregating multiple N_Port logins from servers into one or more external links to the core Fibre Channel fabric. To the upstream FC switch, the FI appears as a host with multiple devices (via NPIV), not as another switch. This simplifies SAN administration and avoids domain ID limitations. FC Switching Mode would make the FI a full-fledged FC switch, which is not the requirement.

A Rollback Workflow is a crucial component for creating resilient automation. It is a separate workflow designed to execute automatically if any task in the primary workflow fails. Its purpose is to reverse the changes made by the successful tasks that ran before the failure, returning the system to its original state and preventing partially completed, failed configurations.

Stretching the Layer 2 broadcast domain (VNI) across both data centers is fundamental. It ensures that the VM's original subnet is available in the destination data center, so it does not need to change its IP address after the move.

Using a distributed anycast gateway, where the same default gateway IP and MAC address exist on all VTEPs in both data centers, is crucial. This ensures that when the VM moves, its default gateway is immediately available locally without requiring any ARP updates, preventing traffic from being black-holed or inefficiently routed back to the original DC.