CCNP DC · vPC+

Nexus: vPC+ on a FabricPath edge

SW3 bundles its two uplinks into Port-channel20 with LACP, one to each leaf. Inside a FabricPath fabric the leaves pair up with vPC+: a vPC domain whose peer-link is a FabricPath port and which is known to the rest of the fabric as one emulated switch-id.

CCNP Data Center track · Cisco Nexus (NX-OS-style) + Cisco · 6 objectives

Open in simulator More CCNP Data Center labs

Objectives

The simulator ticks each one as soon as the network meets it.

On L1 and L2 enable feature lacp and feature vpc, and create vpc domain 10 with the peer-keepalive over mgmt0 (10.255.0.1 ↔ 10.255.0.2, VRF management); give L1 role priority 100.
In the vPC domain set the emulated switch-id: fabricpath switch-id 1000 (the same on both).
Bundle Ethernet1/5-6 into port-channel1, make it switchport mode fabricpath and the vpc peer-link.
Bundle Ethernet1/4 into port-channel20 (LACP), trunk VLAN 10, and make it vpc 20.
show fabricpath route on S1: switch-id 1000 is reached over two next hops (L1 and L2).
Ping PC2 (192.168.10.20) from PC3.

Start this lab

Before you start

vPC+ is vPC inside FabricPath: the peer-link must be a FabricPath port, and both peers announce the hosts behind their vPCs as one emulated switch-id, so the rest of the fabric balances traffic to them over both leaves.

Concept

FabricPath: routing Layer 2 frames between switch-ids

FabricPath replaces spanning tree inside the data-center core. Every switch has a switch-id, and FabricPath IS-IS computes shortest paths between switch-ids over all FabricPath links, so no core link is blocked and equal-cost paths are all used. Frames from classic Ethernet are encapsulated with the source and destination switch-id at the edge.

Concept

vPC: one port-channel across two Nexus switches

A virtual port-channel lets a downstream switch or server bundle links that end on two different Nexus switches. The two switches form a vPC domain: they synchronise over a peer-link and check each other is alive over a separate peer-keepalive. Toward the access switch they act as one LACP partner, so both uplinks forward and spanning tree blocks nothing.