Fabricpath Multicast pakket question

[Sathees1986]

Hey guys im working on a project about fabricpath and i have hardtime with understanding how fabricpath handles the multicast pakkets in following scenario.

Lets say end node send a multicast pakket to all nodes in vlan 10. this pakket will be send bij switch1 to FP1 and FP2 switches. those switches will send that pakket to the multicast root switch. And the root switch will get both pakkets and will send it to all FBswitches who have acces to vlan 10. this mean FP1 and FP2 will also get this pakket. And both will decap the pakket and will see that the pakket is from end node which is source to there edge. because of this FP1 and FP2 will send this pakket agian to the root switch. this mean those pakket will be in a loop. I know that TRILL is solving this problem with have a Appointed forwarder. And i have read that that Fabricpath uses designated forwarder. Nou my question is how is the DF selected?

hoop you guys can help me with understanding this.

Kind regards,

Sathees

[richbarb]

Hello Sathees,

The Fabricpath builds a tree topology schema to forwarding multi destination traffic. The traffic will reach all destinations following the topology that specific vlan belongs to.

Documentation reference:

http://www.cisco.com/en/US/partner/docs/switches/datacenter/nexus5500/sw/fabricpath/6x/fp_n5500_forwarding.html#wp1804761

Regards,

Richard

[Sathees1986]

Hey Richard,

Thx for your reply. I know that Fabricpath uses Multidestination tree to forward multicast packet. but my question was how does Fabricpath prevended loops in scenarios like this.

when End node send a multicast packet to all host in his vlan. Switch1 will get that packet and wil send it to FP1 and FP2. FP1 and FP2 will send that packet to the multidestination tree Root. and the root will send that packet to all Fabricpath switches which can acces that particular VLAN. this means that FP1/FP2 will also get a multicast packet. when they Decap the packet they will see that it does belong to a end node which is connected on theere edge. se both FP1 and FP2 will send the packet back to the Multidestination Root switch. at this point there will be a loop in the network.

how does Fabricpath solve this problem?. TRILL solve it with haveing a Appointed Forwarder.

Hope you can help me out.

Kind regards,

Sathees

[Amit Singh]

Hi Sathees,

If you were to deploy this topology, you have to deploy VPC+ to resolve some of the issues that would run into both for unicast and other traffics. Once you deploy the VPC+ in this topology, your VPC peers will decide the designated forwarder up in the FP cloud and send the multicast packet up to the root. You would manually configure the root-priorities at spine layer for the primary MDT root and all the other FP spine layer switches will become the secondary root. Your VPC+ peers take care of the multicast state exchange to decide the designated forwarder for a multicast group and vPC+ peer device will always use the multicast destination tree rooted on itself to forward multicast traffic.

In a vPC+ pair, one of the switches is elected/configured as the DF. The  DF is the only switch that will forward BUM traffic southbound on the  vPC+ ports towards end stations.

Hope this helps.

Cheers,