The MPLS-OPS Archive

Cell Relay Retreat>MPLS-OPS Archive>month:2003-Aug> msg00091



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

Re: Penultimate Hop Popping

  • From: Rajiv Asati <rajiva@cisco.com>
  • Date: Fri, 29 Aug 2003 15:25:05 -0400
  • Cc: mpls-ops@mplsrc.com
  • Resent-Date: Fri, 29 Aug 2003 16:08:45 -0400
  • To: bhavesh_modi@da-iict.org
  • X-Sender: rajiva@dingdong.cisco.com

Bhavesh,

At 12:54 PM 8/29/2003, Bhavesh Modi wrote:

>Hi all
>
>As RFC 3031 says ( Section 3.16 ), If we do not use Penultimate Hop
>Popping , the egress router will have to do two table lookups even in
>case of only single label in label stack - one for determinig that it
>has to pop the label & second traditional IP lookup.

In a typical case, IP lookup is not needed on the MPLS packets. MPLS 
packets are only subjected to the LFIB lookup.
If the MPLS packet's label corresponds to an outgoing aggregate label, then 
the label stack is ripped off, and then the packet gets subjected to an IP 
lookup.


>But instead of
>only telling to pop the label can't we have one more field which gives
>it next hop?

One more field ? Where ?

LFIB already contains the outgoing interface information and the L2_header 
rewrite in all the cases except the aggregate label case.

>If we have such thing, there is no need of penultimate hop
>popping in case of stack of sigle label atleast.

I am not sure if I follow your premise correctly here.
What are you proposing here ?

Cheers,
Rajiv

>Has any vendor implemented such technique?
>
>-Bhavesh Modi
>MTech-ICT
>DA-IICT.
>Gandhinagar
>
>-------
>The MPLS-OPS Mailing List
>Subscribe/Unsubscribe:  http://www.mplsrc.com/mplsops.shtml
>Archive: http://www.mplsrc.com/mpls-ops_archive.shtml


-------
The MPLS-OPS Mailing List
Subscribe/Unsubscribe:  http://www.mplsrc.com/mplsops.shtml
Archive: http://www.mplsrc.com/mpls-ops_archive.shtml