Owner's Manual

Multi-domain VLT (mVLT)

NOTE: Host NIC teaming from servers or LACP from external uplink switches in the same VLT pair is

supported, but LACP or teaming connectivity across VLT pairs is not supported.

mVLT is a subset of VLT that you can use in a full square mesh network with higher resiliency for link and

node failures and better link utilization. Unlike VLT, mVLT builds two VLT pairs, with each pair hosting two

directly connected switches. The link between the two switches is known as a VLT interconnect (VLTi)

link. Links between the two VLT pairs are considered VLT ports. The software configures the VLT ports

from each switch node to the other VLT pair as a single LAG. LAGs help prevent loops in mVLT

topologies.

To prevent loops in the VLT pairs, the software uses the split horizon and one-copy rule methods. Any

packet from one VLT pair to the other VLT pair is not sent back to the original VLT. Only one copy for any

received flooding, broadcast, or multicast packet from one VLT pair is delivered to the other VLT pair. If a

switch in the VLT pair drops any packets from the VLTi ports to the VLT ports, only the switch that

originated the packets delivers the packets to the other VLT pair and the packets are not duplicated on

the other switch. Use the split horizon method by applying VLT blocking rules to drop any packets from

VLTi ports to VLT ports in all switches that form the mVLT topology. The learned MAC addresses on the

LAG ports acting as VLT ports in one switch in the VLT pair are duplicated in the LAG ports acting as VLT

ports in the other switch in the VLT pair. This replication for learned MAC addresses over VLT ports in the

same VLT pair increases bandwidth efficiency by using multiple paths to the same destination.

Link Failure Scenarios

In an mVLT topology, link failure events are handled differently depending on where the failure occurs. If

at least one VLT port from each switch in the VLT pairs is active, mVLT does not require intervention in

the active switching network. If all VLT ports from one switch in the VLT pair experience a link failure

event, the switch moves all MAC entries from the VLT ports to the VLTi port and reconfigures the other

switch in the VLT pair to remove VLTi to VLT port blocking filters.

If a VLTi link failure occurs, AFC deletes the VLT pair from the switches. The controller configures both

switches in the VLT pair as leaf nodes in another VLT pair. If you have configured NIC teaming or VLT on

legacy switches on any connected servers, the controller disables all edge ports for the secondary VLT

switch.

Node Failure

A single node failure triggers the deletion of the configured VLT pair. The node failure is handled the

same way as a VLTi link fault. Based on the node failure event, the FDB entries are flushed and

reprogrammed.

Multi-Stage mVLT Extension Support

As an extension model of the mVLT topology, you can connect multiple VLT pairs to the existing mVLT

topology. You can attach the new VLT pair to only one existing VLT pair in the topology. The VLT pair can

share multiple VLT pairs.

Supported Technologies and Protocols