The MPLS WG Archive

["James R. Leu" <jleu@laurelnetworks.com>]

On Thu, Jul 20, 2000 at 11:00:35AM +0530, G.B.Naidu wrote:
> 
> Hey,
> 
> What is this LPM Chatur is referring to?

Longest Prefix Match.

> 
> --gb
> 
> On Wed, 19 Jul 2000, James R. Leu wrote:
> 
> > Comment below:
> > 
> > On Wed, Jul 19, 2000 at 06:51:50AM -0700, Chatur sharp wrote:
> > > Hi folks,
> > > 
> > > I have a very simple querry.
> > > 
> > > Consider the scenerio given below
> > > 
> > >           L1       L2
> > >     LSR1 ---- LSR2----LSR3
> > >  
> > > 
> > > The LSR1 and LSR3 are the ingress and egress
> > > respectively for an LSP for FEC F. 
> > > 
> > > Now when LSR1 gets an unlabelled packet, such that 
> > > it maps to FEC F then it will label the packet with
> > > a label L1 and forward it LSR2 who will in turn
> > > forward it to LSR3 with label L2. (LSR2 will swap L1
> > > with L2)
> > > 
> > > 
> > > Now should will LSR3 do an LPM to determine the
> > > interface through which it must forward the packet ?
> > > 
> > > Can it be avoided by choosing appropriate label L2 ?
> > > 
> > > In general, is it always necassary to do LPM at
> > > the egress LSR. ( Assuming no PHP ) ? Put differently
> > > at the end of a LSP, what is generally done - LPM,
> > > or a direct forwarding based on the incoming label.
> > > 
> > > Is there any other forum where I can ask this 
> > > question.
> > 
> > Being that LSR3 allocated the label L2 it can attach whatever
> > meaning it wants to L2.  It all depends on the qualities of the FEC F,
> > the granularity of the label allocation scheme, and how these are tied
> > together in the forwarding plan.
> > 
> > Jim
> > 
> 
> -- 
> Never trust an operating system you don't have sources for. ;-)

-- 
James R. Leu
Software Engineer
Laurel Networks, Inc
jleu@laurelnetworks.com