The MPLS WG Archive

["S. Chkaravorty" <sc12@erols.com>]

True, I didn't mean it that way :-))

Eric Gray wrote:

> S. Chkaravorty,
>
>         Route diversity is not provided per-call,
> for every call, even in the most retentive of
> transmission networks. :-)
>
> --
> Eric Gray
>
> > -----Original Message-----
> > From: S. Chkaravorty [mailto:sc12@erols.com]
> > Sent: Monday, October 23, 2000 5:46 PM
> > To: Kireeti Kompella
> > Cc: jls@research.att.com; ip-optical@lists.bell-labs.com; mpls@UU.NET
> > Subject: Re: [IP-Optical] RE: Optical link bundling. Was Re:
> > DraftMinutes From Pittsburgh
> >
> >
> > Kireeti,
> >
> >
> > Kireeti Kompella wrote:
> >
> > >
> > > Are you (or others) proposing the mechanisms to do this in the
> > > UNI/GMPLS framework?
> > >
> >
> > Why is that so important?
> >
> > >
> > > > The heuristics to simultaneously route a number of
> > circuits simultaneously
> > > > subject to diversity constraints essentially require
> > zillions of calls
> > > > to a Dijkstra algorithm and so don't seem to lend
> > themselves to incorporation
> > > > in the control plane.
> > >
> > > In the IP world, I see the use for simultaneous set up of a small
> > > number of disjoint paths, basically, a primary and one or more
> > > backups for an LSP.  So, this may not be an issue.  What is the
> > > requirement driving having lots of simultaneous diverse paths in
> > > the telephony world?
> > >
> >
> > Very simple.  It is an age old practice in all transmission
> > networks to provide
> > route diversity for failure back-up and peak traffic conditions.
> >
> >