[1818] in Commercialization & Privatization of the Internet
Commercial Use Scenario Part 1
daemon@ATHENA.MIT.EDU (Gordon Cook)
Wed Jan 1 13:13:20 1992
To: com-priv@psi.com
Date: 1 Jan 92 12:29:44 EST (Wed)
From: cook@tmn.com (Gordon Cook)
<<MESSAGE from>> Gordon Cook 01-JAN-92 12:29
cook@tmn
It is my understanding - first that a regional does not have to sign any
of the ANS agreements and - second that it may also sign the connectivity
agreement and "never have to sign either the gateway or cooperative
agreements." I also understand that in talking about customers of the
mid-levels sending commercial traffic across the backbone, ANS is not
challenging in any way traffic of currently connected mid-level customers
designated as ".com" that these customers state is in conformity with NSF
acceptable use policy. Essentially this also means that ANS is not
challenging the status of these connections.
I would surmise that if any mid-level wishes to sign up new customers that
designate themselves as .com, the same conditions will apply. That is the
new customer may use the backbone for traffic that is in conformity with
the AUP of the NSF.
Therefore it would appear to me that much of the current debate is focused
on the ramifications of the following kind of scenario. Let's take a
hypothetical mid-level. Let's say we are talking about a mid- level
having 15 edu gov or org sites at T-1, 20 edu gov or org sites at 56kbs
and 20 edu gov or org sites at eithe 19.2 or 9.6 kbs has say 22 current
customers designated .com. 10 of these customers all linked at T-1,
decide that they wish to start sending corporate spread sheets, sales
projections, and updates to commercial service manuals across the
backbone. This means that the mid-level now has commercial traffic and
must sign either the cooperative agreement with ANS or the gateway
agreement. Lets assume they sign the gateway agreement. Several things
would appear to follow from this chain of events.
1. Presumably 100% of what these 10 com clients send onto the backbone
must now be judged as commercial. Why? Because you don't and presumably
will not in the future actually inspect packets and policy based routing
from what I can gather has not yet arrived (ie is not yet usable).
2. If they send truly enterprise networking commercial data across the
backbone, then they will be sending it to other offices of the same
corporation which will be connected by other mid-levels. Now perhaps the
other offices will be allowed to receive the data with no strings
attached, but one imagines that these offices would then also wish to send
replies. Boom! In such a case we have a situation where 5 or 6 or 7
OTHER mid-levels must if they are to service these companies must also
sign either the gateway agreement or the cooperative agreement with ANS.
True? Or false? It seems that the complexity would increase rather
quickly.
This leaves me wondering the following: how many cases are there now
where multiple offices of the same company are connected to different
points on the network via different mid-levels? I think Apple may be one
via Cerfnet and Barrnet and Nearnet??? Are there others? Probably but I
suspect not many.
The ANS agreements it seems to me are based on the expectation that there
are LOTS of companies looking to use the NSFnet/Ansnet for enterprise
networking. Do we really have solid evidence of this? And why in view of
the kinds of concerns for security of corporate data stated here more than
once by Erik Fair would such companies wish to choose NSFnet/ANSnet for
enterprise networking when they could do TCP/IP commercial wide area
networking through Infolan, or MCI's other service called Vnet if memory
serves me correctly, or through EINet, or Factory America Net, or via
Syncordia's net (Am I correct that this is or will be TCP/IP?). Or by
TCP/IP wan services from AT&T or Sprint? Or probably before long by
services offfered across someone's backbone by Metropolitan Fiber Systems?
The enterprise networking TCP/IP market for 1992 has been pegged at over a
billion dollars. So there is real gold here folks. What I hope that
someone can explain to me is WHY there is a major expectation that with
the complexities of NSFnet/ANSnet - let alone the current animosities a
commercial organization would want to hook up to either ANS or a mid-level
to do purely commercial TCP/IP wide area networking given the alternatives?
Forgeting for the moment the question about a corporation needing multiple
hook-ups to make its use worthwhile would someone from ANS be kind enough
to respond in as much detail as necessary to show the financial impact of
such a scenario on the hypothetical mid-level I have outlined?
A somewhat different scenario is also possible. I shall address this in a
second note to this list. (This will be the scenario where a company like
Dialogue wishes to sell information or oher services to academia.)
