Overview of Multicast Solution

In an ideal multicast solution, the multicast packet is sent only once from a leaf into the spine layer of the network and reduce the replication of flood or multicast packets.

Since the number of spine nodes is less than the number of leaf nodes in a network, the packet replication can be avoided by not sending the same packet multiple times over the same link.

An MDT can be established using BGP and PIM. BGP discovers the VTEPs and PIM establishes the maintains the tree.

There are different types of MDT traffic:

• General or default MDT: The tree is formed from all nodes that participate in VTEP discovery within a domain. Any BUM traffic entering the network is transmitted to all participating leaf nodes and is dropped at egress if it does not apply to the leaf. This may happen if the VNI is not configured at that leaf for user multicast traffic or VLAN does not participate in the c-multicast group.
• VLAN or VNI specific: BUM traffic on the tree is delivered to all members of the VLAN. The user multicast traffic is dropped at egress if the group in the packet is not part of this VLAN at the egress node.
• Group specific: Multicast traffic for a specific group and VLAN can form its own MDT for optimal distribution.

The IP multicast fabric allows user-configured mapping between VNIs and MDTs which supports a default MDT for VLANs or VRFs and VNI specific MDTs, as required. The multicast vxLan tunnels are used to deliver packets. The format of the IP packet is same as a vxLAN packet:

• Destination IP: Group address of the MDT (instead of the remote VTEP)
• Source IP: Sender (leaf node) IP
• VNI: VNI which represents the VLAN in the network

Multicast Distribution for BUM Traffic

The replication of BUM packet must be controlled and reduced. If there are N VTEPs (neighbors) participating in a VLAN, ingress replication sends N copies of the BUM packet.