The IP Over NBMA (ION) Archive

[Albert Manfredi <manfredi@engr05.comsys.rockwell.com>]

Masataka,

mohta@necom830.hpcl.titech.ac.jp wrote:

[ ... ]

> Make UNI/NNI an RSVP clone, make ATM and IP addresses mechanically
> translatable, and so on.

You know what the irony is? It's that RSVP is a UNI/NNI clone. Now you
want to make a UNI/NNI the clone of its clone! But basically I agree
with your intent. I just wonder who should be made the clone of whom.
(To be clear, I think that PNNI is a more complete solution, although I
can't find the promised update at ftp.atmforum.com.)
 
> It'll make ATM swiches eventual IP routers, save packet format.

Of course. Going in this direction, that's what would happen.
 
> The approach should be called "strong peer model".
> 
> Do you like it?

Yes, it's the kind of approach I think will be needed to unkludge the
kludge. But again, the question is whose signaling to use as foundation.

On a somewhat related topic, if we introduce NHRP, why should we make it
different from ATMARP? Why can't ATMARP just be considered a trivial case
of NHRP? In principle, if a host transmits an ATMARP request, why not
respond with a direct ATM address, if the destination is reachable
directly? I mean, whether or not that host thinks it's using RFC-1577 or
Part Deux?

Bert
manfredi@engr05.comsys.rockwell.com