The IP Over NBMA (ION) Archive

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

smirnow@fokus.gmd.de wrote:

[ ... ]

> I'm not an expert, but all documents which I know from Ipsilon and P.Newman
> about IP switch insist on the fact that IP switch has NO ATM software at all
> (no ATM signalling)
> 
> > Tag switching is in Class 2)
> > SITA is in Class 2)
> > ARIS is in Class 1)
> >
> > One of the topics of discussion to this point in time has revolved around flow
> > merging. It would seem to me that Class 1) proposals must modify ATM
> > to make this happen.
> 
> ... or this is your proposal: i.e. you want Ipsilon to modify ATM signalling??

Well, Michail, I also have a hard time keeping all the proposals
straight in my head (probably should dedicate more time to studying
them). Seems to me that the only two proposals which keep ATM signaling
intact are MPOA and Masataka's CSR. All the others are basically trying
to reinvent ATM to provide an easier fit for IP transport.

The philosophical difference between MPOA and CSRs, seems to me, is that
MPOA attempts to make use of Q.2931/2971 for global IP comms over ATM,
while CSRs use ATM signaling only in the local network, sort of in place
of other LAN signaling alternatives. (Forgive me, but it seems a little
like using Boeing 747s only for puddle-jumper local service, although I
do see his motivation.)

Bert
manfredi@engr05.comsys.rockwell.com