[j-nsp] P2MP LSP

Humair Ali humair.s.ali at gmail.com
Tue Jun 29 15:38:45 EDT 2010


Hi David,

Mark is absolutely correct, his example is specific to NG MVPN, although
technically you can also have

L3VPN P2MP, but yeah now best to move to NG MVPN if you can , and get the
benefits of a BGP based core.

Regarding Mark comments that most are using inclusive P-tunnels, we are
using selective P tunnels, although

i think most implementation use inclusive P-tunnels, as it easier to manage
but I personnaly think it add more burden on the network.

Here is also another way of understand how the P2MP is actually build,

AFAIK, think of a network that has


                                  ---> PE4
PE 1  --> P1  ---> P2    ---> PE3
                                  ---> PE5

First, you enable the p2mp in your mpls config for the lsp's

what happened is that when traffic is destined from PE1 to PE4 , it create a
Point to Point LSP with the labels

when a second traffic request, then goes from PE1 to PE3, it creates a 2nd
P2P LSP,again with it's own labels

a 3rd traffic request comes in which is from PE1 to PE5 , it creates a 3rd
P2P LSP, also with it's own labels

The P2MP options that is enable on MPLS, what it does is that the common LSP
path between the 3 endpoints then gets merge and it create a single label.

So in our examples between PE1 and P2 (PE1 + P1 + P2 ), this part of the
path will be merged and will share among them a single label for all 3
destinations,  and then becomes a true P2MP LSP.

HTH

BR

On 29 June 2010 18:21, Mark Tinka <mtinka at globaltransit.net> wrote:

> On Tuesday 29 June 2010 11:43:49 pm David water wrote:
>
> > How the ingress node knows about all the egress Node?
> >  BGP?
>
> Yes, the MCAST-VPN address family ('inet-mvpn' in JUNOS)
> signals the MVPN BGP NLRI either between peers or between
> peers and route reflectors that support this AFI.
>
> > Now ingress node knows about the egress node then tunnel
> >  will be singaled using the RSVP so there must be
> >  interaction between BGP and RSVP right?
> >  If so then what
> >  kind of communication is it?
>
> P-tunnels transport MVPN traffic across a p2mp LSP from
> ingress to egress. The tunnel typically implemented by JUNOS
> for the NG-MVPN infrastructure is MPLS, signaled by RSVP-TE.
>
> The ingress/Sender PE router uses a new BGP attribute called
> PSMI (Provider Multicast Service Interface) to disseminate
> P-tunnel information to the egress/Receiver PE routers. The
> Receiver PE routers then join the P-tunnels, and become
> leaves of it.
>
> I-PMSI P-tunnels (inclusive) forward MVPN data to all PE
> routers in the RSI. S-PMSI P-tunnels (selective) can
> restrict this to only a group of PE routers in the RSI. Both
> options have their pros and cons, although inclusive tunnels
> tend to be more common (I think).
>
> On the control plane side, BGP is used to distribute
> Multicast routing information across the backbone, in
> effect, replacing PIM in the core.
>
> >  Any good link or documents?
> >  If any one has any good document to share that will be
> >  great help, I am looking for control plane communication
> >  between Egress nodes and Ingress node to establish P2MP
> >  LSP.
>
> These are great documents:
>
> http://www.juniper.net/us/en/local/pdf/whitepapers/2000320-
> en.pdf
>
> http://www.juniper.net/us/en/local/pdf/app-notes/3500142-
> en.pdf
>
> Cheers,
>
> Mark.
>
> _______________________________________________
> juniper-nsp mailing list juniper-nsp at puck.nether.net
> https://puck.nether.net/mailman/listinfo/juniper-nsp
>


More information about the juniper-nsp mailing list