IPv6 Multicast over MCT

Multicast over multichassis trunks (MCT) uses two BGP EVPN route types, Join Sync Route and Leave Sync Route, to exchange Layer 2 multicast routes.

Multicast state information is synchronized between the MCT peers using MP-BGP EVPN transport, specifically the Join Sync Route and the Leave Sync Route. Multicast protocol packets are sent on the peer link only when required.

MLD (Multicast Listener Discovery) protocol packets are of three types.

Multicast listener query: General query and Group-specific query
Multicast listener reports: Version 1 and Version 2
Multicast listener done: Version 1 and Version 2

MLD query packet processing

Each EVI is associated with a multicast group ID (MGID) that is BUM-suppressed (Broadcast, unknown unicast and multicast). However, query packets need to be transmitted on an interchassis link (ICL) to address the following scenarios.

The querier connected to only one of the MCT peer switches is the elected querier.
Only one of the peer switches is configured as a querier.
The switch ages out MLD routes if memberships are not confirmed in the timeout interval. Although query packets are received on the MCT peer link, the mrouter port is not learned or considered on that peer link.

MLD membership reports

Traditionally, each peer switch learns about Layer 2 multicast memberships by snooping the MLD membership reports. The membership reports are then flooded on multicast router (mrouter) ports.
For MCT, because an mrouter port is not learned on the peer link, membership reports are not flooded between the peer switches. Peer switches exchange learned routes by using EVPN NLRI messages between Border Gateway Protocol (BGP) peers running on the MCT cluster control VLAN.
The BGP on the peer switch communicates with the MLD routes to the multicast module, which adds and deletes group memberships. The module also allocates or rejects the MGIDs and installs or uninstalls the routes in hardware.
If a general query or group-specific query is received from any port other than a peer link, each peer switch generates a proxy report for the MLD routes learned from EVPN NLRI.

Duplicate MLD query packets on CCEP

If a query is configured on both MCT peers of a member VLAN, duplicate query packets reach clients that are connected to the MCT domain by means of a Cluster Client Edge Port (CCEP).

Multicast modules use inclusive multicast routes of EVPN NLRI to exchange the multicast querier configuration on the member VLANs. The MCT peer that is the DF of the member VLAN is elected as the querier.

Multicast modules handle querier election for the member VLAN when peer links are active or inactive.

MLD listener done

When fast-leave is not configured and an MCT peer receives MLD Listener Done from one of its clients for group G, the switch or router informs other MCT peers about the group-specific query and latency by using the Leave Sync Route. The peer switch, which runs the querier, sends group-specific queries and group queries.

Mrouter synchronization

Mrouter synchronization helps in achieving optimal path selection for unknown multicast traffic and optimal MP-BGP message exchange between MCT peers. The node that is elected multicast DF (designated forwarder) for the VLAN forwards data traffic on CCEP as an mrouter port.

Scalability


Scalability feature	Maximum
MLD snooping-enabled VLANs or bridge domains over MCT	256
IPv6 PIM-enabled VEs over MCT	32
MLD groups over MCT	4000
IPv6 PIM cache routes over MCT	2000

Device support

Extreme Networks supports IPv6 multicast over MCT on the following devices.

ExtremeSwitching SLX 9540
ExtremeRouting SLX 9640