The MPLS-OPS Archive

Cell Relay Retreat>MPLS-OPS Archive>month:2002-Jul> msg00100



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

Re: AW: Penultimate Hop Popping

  • From: Robert Raszuk <raszuk@cisco.com>
  • Date: Wed, 31 Jul 2002 22:14:59 +0200
  • CC: "'roberto'" <roberto_narvaez@yahoo.com>, Aamer Akhter <aakhter@cisco.com>, mpls-ops@mplsrc.com
  • Organization: Signature: http://www.employees.org/~raszuk/sig/
  • Resent-Date: Wed, 31 Jul 2002 17:23:19 -0400
  • To: Hummel Heinrich <Heinrich.Hummel@icn.siemens.de>

HH,

> But is all of this worth the effort?

Yes.

> And is it fair to blame any switch in case it cannot handle these
> top-of-the-art mechanism?

Switches are not required to do PHP. Example: ATM switches.

> These mechanism will only make sense if there is still an IP-destination address in the packet header
> at the egress. They won't make sense in case of VoMPLS,etc...

For non IP packets you usually use outer encapsulation anyway to save
state in the core. I am afraid by VoMPLS you mean Voice over MPLS and I
can't see how this is doable other then on peer to peer basis. VoIPoMPLS
works just fine instead. Is it worth to reinvent the wheel with VoMPLS ?

> They are a burden for any reasonable OAM-concept. 

No, see draft-ietf-mpls-lsp-ping-00.txt for starters.

> They are a burden for any other new concept.

Maybe such a "new concept" requires further work ...

> Wouldn't it be better to give up these two concepts at all ?

No. See if PHP is a problem you may always use explicit-null. And
implmentation of explicit-null support is as easy as implementation of
generic labal switching.

Rgs,
R.



> Hummel Heinrich wrote:
> 
> There is this fantastic savings due to penultimate label popping: No label look-up at the egress.
> 
> And there is this other fantastic savings due to IPv4-Explicit-Null label:
> No label consumption from th egress node's label space.
> Remember, B.Jensen in Yokohama:" All routers shall learn to handle  the IPv4-Explicit-Null label" !
> 
> But is all of this worth the effort? And is it fair to blame any switch in case it cannot handle these
> top-of-the-art mechanism?
> 
> These mechanism will only make sense if there is still an IP-destination address in the packet header
> at the egress. They won't make sense in case of VoMPLS,etc...
> 
> They are a burden for any reasonable OAM-concept. They are a burden for any other new concept.
> At least,they have to be scrutinized whether they jeopardized these two important mechanisms.
> 
> Wouldn't it be better to give up these two concepts at all ?
> 
> Heinrich
> 
> 
> -------
> 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