The MPLS WG Archive

[Eric Gray <eric.gray@sandburst.com>]

Neil,

    Three comments WRT this question:

    1) This is one of the really good reasons why you might use the pipe model -
        you probably would not want to age a customer's packets differently because
        of an internal routing change (or - in your example - mishap);

    2) Unless the LSR is patently incapable of being an intermediate LSR, it will
        almost certainly have stored the current hop-count for whatever Label it is
        presently using for any particular LSP;

    3) If you're alluding to the idea that hop-count may not have been provided, I
        believe this to be very unlikely in the non-TTL capable LSP case.

You wrote:

> Eric....what happens if you decrement at an ingress by N say (N being the
> 'guessed/known' number of hops) and then protection switching occurs such
> that the hops change to M now.....does one have to re-set the ingress
> decrement to the new 'guess/known' number of hops?
>
> regards, Neil
>
> > -----Original Message-----
> > From: Eric Gray [mailto:eric.gray@sandburst.com]
> > Sent: 26 November 2001 20:44
> > To: David Allan
> > Cc: mpls
> > Subject: Re: MIBs and TTL
> >
> >
> > David,
> >
> >     For non-TTL capable links, the TTL should have been decremented at
> > the ingress, rather than the egress.  The ingress - for example - may
> > decrement by one, or by the hop-count.  If these are the choices, the
> > actual decrement at the ingress will depend on the model of the LSP.
> >
> <snipped to end>

--
Eric Gray (mailto:eric.gray@sandburst.com)
http://www.mindspring.com/~ewgray