[2092] in Commercialization & Privatization of the Internet
Are We to Buy Speed for a Few or Connectivity for Many?
daemon@ATHENA.MIT.EDU (Gordon Cook)
Sat Jan 25 15:45:00 1992
To: com-priv@psi.com
Date: 25 Jan 92 15:32:53 EST (Sat)
From: cook@tmn.com (Gordon Cook)
<<MESSAGE from>> Gordon Cook 25-JAN-92 15:32
cook@tmn
~rRob Horn horn%hydra@polaroid.com says:
>...They don't want NSF picking their vendor.
>It appears to me that the commercial offerings for IP network >services
are now varied enough and strong enough that NSF >should change its
funding methods ASAP. It can now switch >to a model similar to that used
for computers, where the >researcher is granted real money. NSF allows
researchers to >choose the vendor, and only in special situations is the
>vendor or computer selected or provided by NSF.
Although I would change his model from researcher to institution, I think
Rob raises some excellent points. Many in the community apparently don't
see their interests being well served by the emphasis on the gigabit
network because the changes that will result from attaching everyone to
one backbone that does gigabits are now becoming clearer to the community
as a whole.
Hypothesis: Emphasis on speed leads to emphasis on a single backbone. If
we admit the limitations that lie behind the emphasis on speed point of
view, then perhaps we can see more clearly a way to make alternative
choices.
Let me explain what I mean:
While Robert Lucky didn't use the specific word hype in his address to the
Gigabit symposium in DC last July 16th..... his message was clear: who
needs gigabits? No one. And as a high level official from an IXC told me
recently: a potential tragedy of NREN is that telecom's ability to help
K-12 may be hindered by creating the popular mis-impression that if you
can't get gigabits to the classroom, it somehow just isn't worh it. Like
it or not, Congress has been sold the idea of the gigabit network as the
solution to problems of competitiveness and better research and education.
This means that unless re-education is accomplished, the emphasis is
likely to remain on a single federally funded high speed backbone.
The model that have been out there for emulation are the backbones of the
mission agencies. (DARPA, NASA, DOE, & NSF.) In 1990 Senator Bennet
Johnston was pushing very hard to have DOE and its backbone chosen as the
lead agency for NREN. I have heard some say that going very quickly to
T-3 on its backbone helped to make sure that NSF would be chosen as lead
NREN agency. NREN afterall has been sold as a march to the giagbit
network. (Now before I get flammed let me say I KNOW that there are a lot
of interesting applications in the high tens of megabits per user and that
if you aggregate the traffic of many users over a single backbone, you
will eventually reach a gigabit - never mind however that there is a world
of technology difference and cost difference between delivering an
aggregate and a clear channel gigabit. And never mind that the question
could be raised as to whether emphasis on making the network more user
friendly at current speeds and increasing the percentages of current users
might do more for the state of american research and education that moving
on toward a blindly fast, very expensive network used by a tiny elite.)
I wonder if a strong case could be made for continuing the gigabit
testbeds as a source of new technology development while decoupling the
NSF and the US Government from the idea that one lead agency must supply
an ultra high speed backbone regardless of whether the marketplace seems
to want it or not?
Yes, Rob is right, letting institutions decide from whom they will buy
their network connectivity, could indeed be an idea whose time has come.
However there could be a complicating factor in this scenario. The extent
to which the mid-levels are dependent on NSF support either for backbone
access, or in other ways, is something apparently know only to each mid-
level network director and the corresponding grant giver at the NSF.
Perhaps someone should take on the responsibility for clarifying this and
developing a two year time table to phase down these grants in order not
to pull the rug out from under otherwise deserving mid-levels?
It would also be instructive and helpful to policy makers to publish data
showing which institutions are currently paying the full cost of their
access to the network. It might after all give a clearer idea of what the
current subsidies are buying.
Indeed would it not be correct to assert that 40 or 50 or even 60% of
institutions attached to the network are currently paying their own way
and if network interconnectivity were guaranteed would be free right now
to chose their own backbone provider?
Is it a matter first
of deciding whether the gov't shall continue to fund a
backbone
and second, if the answer is no,
of deciding how to wind down support for that backbone
in such a way as to not topple those entities (such as
mid-levels) that had depended on it?
