The IP Over NBMA (ION) Archive

[Masataka Ohta <mohta@necom830.hpcl.titech.ac.jp>]

Bert;

> Masataka, OK, so you don't see the need to subdivide any given campus
> into multiple IP subnets.

Wrong. There are a lot of reasons to subdivide campuses into multiple
IP subnets, none of which needs cut-through.

> But there is surely a need to subdivide the
> local nets all over a continent into IP subnets, for routing purposes.

"routing purposes"?

I'm afraid you are not routing in the senario below.

> If I have physically large ATM clouds,

I think you mean geographically large.

> but not populated with an
> enormous number of stations, such as a large corporation with offices
> located here and there might have, then for the sake of IP routing I
> will have many LISs.

It seems to me that you assume LISes are not logical and confined
in offices.

Otherwise, you can put everything into a single LIS and use a single
ATMARP server, which is as fault-tolerant as single NHSes over inter-
office LISes.

So, let's assume that we don't have SCSP, on which we are working
though, and LISes are not logical.

> But the people in this corporation can all
> intercommunicate directly over the ATM cloud, without going through IP
> routers, so it is to their advantage to do so for more consistent
> quality comms.

Not at all. As the offices are connected through the egress IP routers,
there is no inefficient path exists over IP routers and there is no
reason to have cut-through.

						Masataka Ohta