300-610 DCID: Designing Cisco Data Center InfrastructureDemo

An OOB management network provides a dedicated Layer 3 management path, which supports device monitoring. vPC peer-keepalive messages require Layer 3 connectivity; Cisco NX-OS uses the management ports and management VRF by default when a separate keepalive VRF is not configured.

Cisco NX-OS does not support direct communication between VDCs on the same physical device. Connectivity can be provided by routing each VDC through an external Layer 3 device or by making a physical cross-connection between a port allocated to one VDC and a port allocated to the other VDC.

OTV maintains site independence by preserving Layer 3 failure boundaries, so failures do not propagate past an OTV edge device. It also filters First Hop Redundancy Protocol traffic, including HSRP hello messages, across the overlay, allowing HSRP to remain local to each data center.

A SAN port channel aggregates multiple physical Fibre Channel interfaces into one logical interface, providing load balancing and link redundancy in a Fibre Channel fabric. If one member link fails, traffic continues over the remaining links.

Computer-aided design workloads require graphics processing capability for accelerated rendering and visualization, so the selected Cisco HyperFlex platform must support GPUs.

In a vHBA redundancy template pair, the primary template defines the shared configuration, while the secondary template inherits the primary template’s attributes. The Initial and Updating settings instead determine whether later template changes update vHBA instances.

FCoE requires a 2158-byte MTU in its network QoS system class. The configured 2000-byte MTU for class-fcoe is too small for the supported Fibre Channel/FCoE payload, so maximum-size FCoE frames cannot be forwarded correctly. Cisco documents 2158 bytes as the FCoE payload MTU across the switch.

Cisco NX-OS natively provides a Python interpreter at the CLI prompt, allowing Python scripts to run directly on the device. Ansible, Puppet, and Chef are external automation tools.

Fibre Channel connectivity is normally provided through redundant paths on separate fabrics, with host or hypervisor multipathing software (such as MPIO) detecting a failed path and redirecting I/O to the surviving path. Cisco UCS fabric failover is typically used for Ethernet vNICs rather than Fibre Channel HBAs. Cisco Unified Computing System Solution Overview

A bridge-domain subnet provides the Cisco ACI fabric default-gateway function only when unicast routing is enabled on that bridge domain. Routing is enabled on BD1, while the Web servers are attached through BD2, so BD2 cannot provide their routed gateway. The NAT appliance must provide the default gateway for those servers.

Cisco vPC+ extends vPC to interoperate between a Classical Ethernet vPC domain and a FabricPath cloud. It makes the FabricPath-enabled peers form a single vPC+ virtual switch as viewed by the rest of the FabricPath network.

iSCSI nodes are identified with either an iSCSI Qualified Name (IQN) or an EUI-format name. WWPN and WWN identify Fibre Channel entities, and an OUI alone is not a complete iSCSI node naming format.

Extended Receiver Ready (ER_RDY) divides a Fibre Channel ISL into virtual links with independently assigned buffer-to-buffer credits. Congestion isolation can place traffic for a slow device onto a lower-priority virtual link, protecting normal traffic. A device can be designated as slow through Port Monitor, which enables that congestion-isolation treatment.

An iSCSI node is identified by a globally unique, persistent iSCSI name rather than by a network address. An IQN uses a domain/host name to identify the naming authority and can include an identifying tag, such as a host identifier, to make the participant’s name unique. Port numbers identify network endpoints, preshared keys authenticate access, and the name type designator identifies only the naming format.

VRF-Lite uses separate VRF routing tables and address spaces, allowing overlapping IP addresses in different VRFs. A VRF can use either physical ports or logical interfaces, such as VLAN SVIs. A Layer 3 interface is assigned to only one VRF, and VRF-Lite does not use MPLS to support VPNs.

Erasing the configuration on both fabric interconnects clears their existing configuration and reboots them to initial setup, where each can be configured as a standalone device. The connect local-management command only enters the local-management CLI; it does not change cluster membership.

Dedicated compute nodes allow Cisco HyperFlex clusters to scale compute resources independently of storage. This lets desktop virtualization environments add capacity for additional virtual machines without purchasing unnecessary storage capacity, providing cost-effective growth.

Cisco UCS Fabric Interconnect end-host mode is N Port Virtualization (NPV) mode, which presents connected vHBAs as N-ports to the upstream SAN. The Cisco MDS switch must enable N Port ID Virtualization (NPIV) to allow multiple N-port identifiers over the shared physical port-channel connection.

Cross-connected with pinning is a supported FCoE uplink design. The Fabric Interconnects and Nexus switches are cross-connected without being port-channeled together, allowing unified uplinks while retaining the established network design.