The MPLS WG Archive

Cell Relay Retreat>MPLS WG Archive>month:2006-Mar> msg00009



[Date Prev][Date Next][Thread Prev][Thread Next]  
  [Date Index][Thread Index][Author Index][Subject Index]

[mpls] I-D ACTION:draft-ietf-mpls-ldp-p2mp-00.txt

  • From: HeinerHummel@aol.com
  • Date: Sat, 4 Mar 2006 05:31:56 EST
  • Cc: mpls@lists.ietf.org
  • X-Spam-Flag: NO

Ice,
 
 It is a MP2MP LSP from the point of view of the leafs, there is an
upstream path to send to and a downstream path to receive from.
 
HH>:I guess, you wanted to say: there is an upstream path to receive from and a downstream path to send to.
 
If you
look at the internals, it is a P2MP LSP from the root to all the leafs.
There is a P2P LSP from the leaf to the next upstream LSR until you
reach the root. While MPLS packets are forwarded upstream, they are
allowed to go down the P2MP LSP if there are receivers.
 
HH>: Look at the mp2mp-LSP as a whole. Data is sent from any particular leaf node to all other leaf nodes inside the LSP by forwarding the data at any junction to all adjacent neighbor nodes (and nodes of the LSP) except the one from where it is received. Depending on which leaf is the data originator, any link or trail inside the LSP is once upstream and once downstream. It is this basic understanding I am after. Accordingly the label mapping process could be organized. But I have my doubts that it will. Let's see what the respective draft will look like.
 
Heiner
 
 


Thx,

Ice.
 
 
_______________________________________________
mpls mailing list
mpls@lists.ietf.org
https://www1.ietf.org/mailman/listinfo/mpls