RE: Draft minutes from Tove: draft-dachille-inter-region-path-setup-04.txt

["Vishal Sharma" <v.sharma@ieee.org>]

Adrian,

> -----Original Message-----
> From: Adrian Farrel [mailto:adrian@olddog.co.uk]
> Sent: Tuesday, December 07, 2004 4:25 AM
> To: v.sharma@ieee.org; Ugo Monaco
> Cc: ccamp@ops.ietf.org; Alessio D'Achille; Daniele Alì; Marco Listanti;
> Tove Madsen
> Subject: Re: Draft minutes from Tove:
> draft-dachille-inter-region-path-setup-04.txt
>
>
>
>
> > And, I should add, that the only protocol "extension" that
> > draft-dachille requires is an ARO object, which is cloned
> > after the ERO object, and can simply be "flipped" to form
> > the ERO object of the second (diverse) path when setting
> > up that path.
> > (As was illustrated in the fig. on slide 3 of the D.C.
> > presentation.)
>
> We're in agreement.
> There is a protocol extension and consequent increase in signaling load.
> It is not a substantial increase.

I'm not sure why a protocol extension should lead necessarily to an increase
in signalling load?

Is the load you're talking about simply that an additional
ARO object needs to be carried during the setting up of the first
of the two LSPs? Or, do you mean something else?

Then again, if one is trying to set up two paths, clearly some
additional signaling will be required (beyond that required
for one path) in any approach.

The main benefit of the ARO approach is the fact that it achieves
this with minimal modifications to the signaling protocol and with
minimal overhead, and without requiring new protocols to be defined
or new uses of existing protocols to be defined.

-Vishal