[114] in Commercialization & Privatization of the Internet
Re: Regionals and commercial providors (was: Re: Is SUNFLASH...)
daemon@ATHENA.MIT.EDU (Martin Lee Schoffstall)
Sun Nov 11 22:14:52 1990
To: sob@tmc.edu (Stan Barber)
Cc: com-priv@psi.com
In-Reply-To: Your message of Sun, 11 Nov 90 19:53:18 -0600.
Date: Sun, 11 Nov 90 20:51:39 -0500
From: "Martin Lee Schoffstall" <schoff@psi.com>
> I disagree that all regionals are basically passive with respect to all the
> political/economic developments concerning NREN (or whatever the future
> of government-sponsorted networking will be called). Some regionals (and
> certainly SesquiNET/THEnet) are taking an active role is providing mid-level
> connectivity to as many "national" backbones as possible. [SesquiNET/THEnet is
> connected to the NSFnet, NSN and ESnet currently.] This means that one
> connection to SesquiNET/THEnet in Texas connects you to multiple long-haul net
> without dragging a line in to the local site for each one.
ESNET/NSFNET/NSN are bad motivating examples. All represent government
supported networks which are free to the connector, have policy restrictions,
and have a value added of X at least to subscribers of TexNet. My guess
is that they represent some kind of bit-dump to TexNet subscribers. There
is no users group that represents the campuses to them, you can't travel
around the country and log into terminal servers that they provide, etc.,
TexNet as a retail operation in Texas probably provides all of those
things and more, (I'll bet PSINet has better barbecues for our Summer
users group meeting though).
I'd put TexNet's value added at 2*X to the subscribers of TexNet.
> I believe this is a good thing economically for the local sites, non-profit
> and for-profit.
I'll bet that the restrictions are stifeling at times to the commercial
people.
> Now, one may argue that the NSFnet three layer model is wrong and that the
> ARPAnet model is better [Am I correct in saying that the PSInet architecture
> is basically like the ARPAnet (i.e. direct connections from the provider to
> the end-user and the provider takes are of all the details of how packets
> traverse from one end-user to the other)?] Given the developments here in
> Texas, I find it hard to believe that we (end-users in Texas) would be
> better served by a provider that follows the ARPAnet model (as described).
The ARPANET model is exactly like the NSFNet model, sole source contract
to a provider not responsible to the real customers that they connect
to, using proprietary/custom software and hardware technology, that
even if you could buy you probably wouldn't want to.
The PSINet model is probably just like the TexNet model, contract
with a site, provide value added service, and be responsible to
the customer.
>
> Now, either PSI (or its competitors) can fight the NSF three layer model
> or they can work with it. What I am asking is what PSI (and the other
> providers) want to do. What you seem to be saying is that the regionals
> are doomed or don't care and PSI will not try to work with the three layer
> model because it will be out the door anyway in October of 1992.
In 1985 some of the people that are principals in PSI, helped create
the model, and in 1987 saw the model get tossed. There has been no clear
three layer model on the NSFNet for years, despite much persuasion
and strong arming. It is an illusion, like "policy based routing".
com-priv@psi.com represents a forum where illusions are disspelled,
and the hard issues are discussed openly instead of in smoke filled
rooms. Now if I could only talk Kent England into being the "Dr. Ruth"
for the forum we'd be set.
Marty
