Okay, just after sending that I think I read an explanation that makes
sense and corrects some problems I had. It appears that the PE routers
participating in the MDT join the MDT group, which is specific to an mVPN.
That means on the P/PE side of the network, they are signaling to the
internal RP that they want to join that *,G. When a multicast packet
arrives on a PE from a CE, that packet gets encapsulated in a GRE packet
with the MDT group as the destination address. That flows toward the
provider RP. All of the other participating PEs have joined the MDT group,
so they receive that multicast packet. Once they receive it, they
de-encapsulate it and place it onto the vrf with the associated MDT group
configured on it.
Is that about right? There are still some parts to this that are confusing,
but I think it's starting to make more sense.
John
On Tue, Jan 8, 2013 at 8:40 PM, John Neiberger <jneiberger_at_gmail.com> wrote:
> I don't know why, but this seems to be a subject that just doesn't make
> sense to me. I'm having trouble visualizing the process and I don't yet
> understand all of the steps necessary to configure mVPN.
>
> One thing I'm not entirely sure of is why we would even configure this in
> the first place. My only thought is that it is a service to provide to
> customers and it saves them from having to configure GRE tunnels between
> their sites since regular L3VPN wouldn't support multicast. I don't see why
> they couldn't just configure GRE tunnels between their CE routers and run
> PIM on them. I guess it would work, but the idea of being a service
> provider is to offload some of that headache, right?
>
> Okay, to the technical stuff. I'll walk through the process as I
> understand it and you'll see where I'm getting lost. Let's assume that the
> basic L3VPN config is done and working.
>
> 1. PIM SM is enabled on P/PE routers and an RP is selected.
> 2. PIM SM is configured in the customer network and will show as a PIM
> neighbor inside their VRF on our PE router.
> 3. The MDT address is configured in the VRF.
> 4. The ipv4 mdt address family is configured in BGP with the same peers as
> in the VPNv4 AF.
>
> Here's where I get confused. At some point in this process, the PE routers
> join the MDT group. They also form GRE tunnels between their loopbacks.
> This seems to be an automatic process. I don't understand the need for the
> MDT group. If the PE routers already have GRE tunnels between them, PIM
> could run across them with no problem as-is. What function does the MDT
> perform and what benefit does it provide?
>
> Another question is this: what actually triggers the building of those GRE
> tunnels? Is that a function of enabling the ipv4 mdt address family?
>
> Please un-confuse me. If anyone has any helpful pointers or even old blog
> posts, please point the way.
>
> Thanks!
> John
Blogs and organic groups at http://www.ccie.net
Received on Tue Jan 08 2013 - 20:48:26 ART
This archive was generated by hypermail 2.2.0 : Sun Feb 03 2013 - 16:27:17 ART