RE: FabricPath & PIM-SM Interoperation (OT)

Multicast senders or receives couldn't be "inside" the FabricPath per-se, they have to be on Classical Ethernet ports. Is your question about the control plane signaling or the data plane flow? For the control plane the PIM Join messages are multicast, so they should follow the multicast multidestination tree in the Fabric Path core. PIM Register is unicast so it should follow the normal unicast path through the FP core based on the MAC address of the RPF neighbor for the RP.

In reality I think that multicast was a small afterthought for a lot of these DC specific technologies like FabricPath and vPC. ECMP with multicast is problematic to begin with though, so with FabricPath their workaround was just to build 2 multidestination trees, and then hash between the two of them. Remember with vPC, the Peer-Link *is* in the data plane for all multicast flows, not just the control plane. I'm assuming this is the same for vPC+ with FabricPath, because there's no other way to account for receivers on orphan ports.

You're right though, the docs are sparse (no pun intended). The only things I could find on the external documentation was these two:

Cisco FabricPath Design Guide: Using FabricPath with an Aggregation and Access Topology - Multicast Routing Considerations
http://www.cisco.com/en/US/prod/collateral/switches/ps9441/ps9670/guide_c07-690079.html#wp9000345

FabricPath Forwarding - Forwarding Multicast Packets
http://www.cisco.com/en/US/docs/switches/datacenter/sw/5_x/nx-os/fabricpath/configuration/guide/fp_forwarding.html#wp1804761

Brian McGahan, 4 x CCIE #8593 (R&S/SP/SC/DC), CCDE #2013::13
bmcgahan_at_INE.com
 
Internetwork Expert, Inc.
http://www.INE.com

-----Original Message-----
From: nobody_at_groupstudy.com [mailto:nobody_at_groupstudy.com] On Behalf Of Yuri Bank
Sent: Wednesday, September 18, 2013 1:59 PM
To: Cisco certification
Subject: FabricPath & PIM-SM Interoperation (OT)

Hello everyone,

I posted this on c-nsp, and really didn't get the kind of answers I was looking for. Hopefully there is someone out there that can dive a little deeper into this topic.

Basically, I would like to know the details on how a switch running FP (Nexus 7k or 5k) deals with multicast sources/listeners that are inside a FP domain, when it is running PIM-SM on a L3 interface(SVI) that is connected to some other part of the network ( that isn't running FP).

For example, my understanding is that a FabricPath edge switch will flood GM-LSPs if it has an interested listener for a multicast group. When the FP switch that is also running PIM-SM on its L3 interface sees this, it would make sense for it to send a PIM-Join message to its neighbor, toward the RP. The same thing should happen with multicast sources, and PIM Register messages.

Does anyone have experience with this? Is my assumption correct here? Any details would be appreciated, as the Cisco documentation on this topic is quite limited.

Thanks,

YuriB.

Blogs and organic groups at http://www.ccie.net
Received on Wed Sep 18 2013 - 15:08:22 ART