Administrator Guide
If the VLT node elected as the designated router fails, trac loss occurs until another VLT node is elected the designated router.
VLT Multicast
VLT multicast provides multiple alternate paths for resiliency against link and node failures.
This feature supports inter-server multicast communication between top-of-rack (ToR) switches using an inter-VLAN Layer 3 routing
protocol (for example, PIM, IS-IS, or OSPF). It also provides trac resiliency during multicast routing convergence after failure
without disrupting or altering multicast routing behavior.
Layer 2 protocols from the ToR to the server are intra-rack and inter-rack. No spanning tree is required, but interoperability with
spanning trees at the aggregation layer is supported. Communication is active-active, with no blocked links. MAC tables are
synchronized between VLT nodes for bridging and IGMP snooping can be enabled.
VLT multicast is also scalable, so additional racks can be implemented in an existing system to provide additional resiliency and
resources to accommodate an increased need or to anticipate future growth. You can enable VLT unicast and multicast routing
across multiple congurations using VLT links. Protocols such as BGP, OSPF, and PIM are compatible with VLT multicast.
Important Points to Remember
• You cannot congure a VLT node as a rendezvous point (RP), but any PIM-SM compatible VLT node can serve as a designated
router (DR).
• You can only use one spanned VLAN from a PIM-enabled VLT node to an external neighboring PIM router.
• If you connect multiple spanned VLANs to a PIM neighbor, or if both spanned and non-spanned VLANs can access the PIM
neighbor, ECMP can cause the PIM protocol running on each VLT peer node to choose a dierent VLAN or IP route to reach the
PIM neighbor. This can result in issues with multicast route syncing between peers.
• Both VLT peers require symmetric Layer 2 and Layer 3 congurations on both VLT peers for any spanned VLAN.
• For optimal performance, congure the VLT VLAN routing metrics to prefer VLT VLAN interfaces over non-VLT VLAN interfaces.
• When using factory default settings on a new switch deployed as a VLT node, packet loss may occur due to the requirement that
all ports must be open.
• You can enable ECMP on VLT nodes using VLT unicast; however, ECMP is not compatible on VLT nodes using VLT multicast. You
must use a single VLAN.
Conguring VLT Multicast
To enable and congure VLT multicast, follow these steps.
1. Enable VLT on a switch, then congure a VLT domain and enter VLT-domain conguration mode.
CONFIGURATION mode
vlt domain domain-id
2. Enable peer-routing.
VLT DOMAIN mode
peer-routing
3. Congure the multicast peer-routing timeout.
VLT DOMAIN mode
multicast peer-routing—timeout value
value: Specify a value (in seconds) from 1 to 1200.
4. Congure a PIM-SM compatible VLT node as a designated router (DR). For more information, refer to Conguring a Designated
Router.
5. Congure a PIM-enabled external neighboring router as a rendezvous point (RP). For more information, refer to Conguring a
Static Rendezvous Point.
6. Congure the VLT VLAN routing metrics to prefer VLT VLAN interfaces over non-VLT VLAN interfaces. For more information,
refer to Classify Trac.
Virtual Link Trunking (VLT)
835