[10518] in Commercialization & Privatization of the Internet
Re: Chronicle of higher ed & the New Math of NSFnet
daemon@ATHENA.MIT.EDU (Hans-Werner Braun)
Sat Feb 26 01:07:19 1994
From: hwb@upeksa.sdsc.edu (Hans-Werner Braun)
To: cook@path.net (Gordon Cook)
Date: Fri, 25 Feb 94 9:31:55 PST
Cc: com-priv@psi.com
In-Reply-To: <9402250534.aa06288@pandora.sf.ca.us>; from "Gordon Cook" at Feb 25, 94 5:34 am
Gordon:
> LORD! The Chronicle of Higher Education *STILL* doesn't get it! Apparently
> they still are a shill for the NSF is getting out of the NSFnet business.
> (Scale back. support..... Got that people? Uh Huh!)
I am sorry, but it appears to me that you are the one who does not get
it. Did you really read the solicitation, listen to NSF and others?
1. The vBNS is *NOT* the NSFNET+. It is *NOT* general purpose
infrastructure. It is *NOT* for Gordon Cook's com-priv email. It is
*NOT* for the aggregation of 10000 flows simultaneously of little
applications. It is a high end entity of non-general-purpose nature to
research and develop applications that require high amounts of
bandwidth for themselves and to use the supercomputer centers and
others as guinea pigs or facilitators or whatever you want to call it
for this. This is a *new* thing, uncomparable to the NSFNET backbone
and essentially a research and development project. As such, it has
nothing to do with commercialization+privatization.
2. The strategic direction for general purpose infrastructure is to
move things over to commercial providers, and perhaps buy some *services*
here and there and as need be. This is only for NSF, some other
agencies plan on own services/infrastructure for their needs.
3. Because the will not be a general purpose NSF funded cook-email backbone
infrastructure any more, NSFNET or otherwise, the NSF is undertaking
steps to facilitate the changes in landscape. As such:
. NSF/DNCRI is planning on giving some of "their children," i.e.,
regional/mid-level networks, some funding to facilitate them getting
used to a mode of buying interconnectivity. The funding is something
like startup funding, and will decrease over time.
. NSF/DNCRI is collaborating with the general purpose infrastructure
by fostering general purpose network interconnection points, major
highway interconnects called NAPs. This should help avoid isolatism
and elitism among service providers, as a NAP service provider has
to sell services to paying customers.
. NSF/DNCRI recognizes that the Internet infrastructure and its
system level (or global interconnectivity or whatever you like to call
it) is not quite mature enough. Particularly few if any take a systems
approach relative to operational stability at a global scale. The
typical mode of operation is that a service provider draws The
Internet Universe centric around the own network, perhaps just
considering the next door neighbors. Few consider the population a
block or few away. At the first order of approximation this can be
argued as a routing problem (yes, yes, yes, that simplifies it a
little, but will have to do here), therefore NSF/DNCRI is taking the
Routing Arbitration awards very seriously. Personally, I think it is a
great thing that NSF awards it to two places, one with a strong
history in research, the other with a strong history in operations,
and pushing them to collaborate. I could imagine that, relative to the
systems level view, the RA is more important than the NAP and vBNS
award (as the latter two are of more local relevance, especially with
a GIX already in existence). The regional interconnectivities for a
while are quite critical though, to avoid major changes too quickly.
So, Gordon, there is really nothing left in these awards that really
resemble much of the NSFNET backbone. The real estate is moving (and
had been moving for years) to commercial providers. While you are
getting what you seem to have been fighting for for years (NSF's
abandoning of general purpose infrastructure) you'll likely in the
future also have to pay the price (like little bills popping up here
and there from commercial providers that you send your email across).
If you are out for adventure, you could also try to apply another
metrics, like how much NSF pays for delivered unit of service (like
packet, byte, net, host, user, ...) and draw a function over time.
Hope your tools can do log scaling.
Hans-Werner
