The MPLS WG Archive

Cell Relay Retreat>MPLS WG Archive>month:2003-Apr> msg00035



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

[PWE3] MPLS PID

  • From: Curtis Villamizar <curtis@fictitious.org>
  • Date: Wed, 02 Apr 2003 09:34:06 -0500
  • cc: Shahram Davari <Shahram_Davari@pmc-sierra.com>, "'Lloyd Wood'" <L.Wood@eim.surrey.ac.uk>, pwe3@ietf.org, mpls@UU.NET


In message <3E8A0542.40603@level3.net>, Luca Martini writes:
> 
> 
> Shahram Davari wrote:
> > I agree with Lloyd. Sniffing IP packets in the middle of an MPLS network is
>  fundamentally wrong, and is impossible to do for all ECMP implementations su
> ch as the one that I described
> > in earlier emails in which both the first nibble and the CRC are used to de
> tect IP.
> > 
> Shahram,
> It might be fundamentally wrong but this is deployed today , and we run our 
> hole network based on this little point. It seems to work fine for a network 
> that transport 90gb/s to 100Gb/s of actual traffic.
> 
> So if we put the theory, and layer arguments aside , what is the issue with 
> determining that a particular flow is not IP ( assuming IP = 99% of traffic )
>  ?
> 
> Luca
> 
> It might be fundamentally wrong
> 
> 
> > The best way to solve this problem is to do ECMP only based on label stack.
>  Doing so should be
> > easier than introducing new standard.
> > 
> > -Shahram


Solving real world problems is never fundamentally wrong.  If all
traffic is know to be either IP or contain a control word then there
is no chance of reorder so its not wrong at all.

Curtis