The IP Over NBMA (ION) Archive

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

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

[ ... ]

> I don't assume RSVP. Any resource reservation scheme including ST2
> is not excluded. Any service class including "best effort" is not
> excluded. Proxy reservation by some intermediate router is not
> excluded.
> 
> It may be a good idea for INT-SERV WG to define "best effort"
> service class. Then, we may be able to replace IFMP with RSVP
> (by adding RSVP fields for encapsulation format etc.).
> 
> > The Ipsilon approach does not apply because it does not interoperate with
> > ATM.
> 
> Wrong.

Actually, I agreed with Joel on that point. My only puzzlement (?) was
that I didn't think CSRs used MARS either.
 
> ATM with some light weight signaling such as RSVP, Ipsilon's or
> Toshiba's is still ATM.
> 
> ATM does not mean Q.2931 signaling.

Whoa. Maybe "cell switching" doesn't mean Q.2931, but ATM _certainly_
means Q.2931.
 
> > It does not work with MARS or NHRP or RFC 1577.
> 
> Isn't LANE not ATM because it does not work with MARS or NHRP
> or RFC 1577?

LANE does use Q.2931, though. That's the key. The other things you list
all use Q.2931. Hence, by def, they are layered on top of something
called ATM.

Point being, in order to make something based on Q.2931 signaling run IP
(or other common LAN protocols), you need MARS/RFC-1577/MPOA/whatever.
If Q.2931 gets upgraded so you don't need those acronyms to run IP,
fine. But if you simply get rid of Q.2931 and replace it with IP
schemes, then you're left with cell-switched IP (hence, "CSR"). Not ATM.

Bert
manfredi@engr05.comsys.rockwell.com
(soon to be .boeing.com)