NSF and the commercial networks -- I asked Rick about the noise surrounding the NAP procurement as h

[David Farber <>]

From: rick () uunet uu net (Rick Adams)
Date: Wed, 9 Mar 1994 18:28:01 -0500 (EST)




The controversy regarding the NSF solicitation stems from a single
fundamental flaw - NSF's attempt to procure services under a
co-operative agreement instead of the more appropriate contract.


The NAP section of the solicitation is probably the strangest. It is
technically, politically and financially flawed.


The NAP award provides for 4 NAPs - Washington, DC, New York City,
Chicago and San Francisco Bay area. Examining these on a city by city
basis proves illuminating.


The Washington DC NAP is undoubtably the most perverse case . First,
there already exists an active, successful, non-federal funded "NAP"
called MAE-EAST.  MAE Stands for Metropolitan Area Ethernet.
Participants are interconnected at 10 MBPs speeds. It works. It's
available now. It costs $2000 per month per participant.  It's provided
by a neutral by pass carrier called MFS.


Strangely, NSF has chosen to pay the same MFS to create an NSF funded
NAP in the same area to interconnect the existing participants. Current
pricing estimates for the NSF sponsored NAP are about 75% MORE
EXPENSIVE that the existing facility. There is no technical benefit to
the participants from the "new" NAP at a significantly higher price.
(Of course NSF has a list of what it considers advantages, but none of
the MAE EAST participants seem to agree with them.)


The New York City NAP is also questionable. There is no technical merit
to putting a NAP in New York City. There aren't even any regional
networks there to attempt to subsidize. There IS an advantage to a New
York City NAP if you happen to be a large phone company. All of the
International Circuits from Europe make landfall in NYC. Obviously,
it's cheaper for a phone company to drop them there. However, if you
are a mere customer of the phone companies, they provide a drop AT NO
EXTRA CHARGE in any major city on the East Coast from Boston to
Atlanta. Yes, that includes Washington, DC where an existing "NAP" is
in place and where most of the European connections already terminate.


Chicago is merely pointless. Its sole apparent function is to subsidize
the few regional carriers in the mid-west. It has no technical merit.
Give the regionals a check instead of cloaking the subsidy in a NAP.


The Bay Area actually has technical merit. Unfortunately, there is no
requirement to waste government money providing it. There has been an
effort to create a Bay Area "NAP" for about 6 months. It keeps failing
because Pac Bell has been unable to deliver reasonably priced high
speed SMDS service.  Interestingly, the NSF awardee is Bellcore/Pac
Bell. Given that Bellcore invented SMDS, want to bet that the Bay Area
NAP is built out of that same SMDS technology currently priced out of
reach for most providers.  MAE-West will exist without NSF funding or
interference if Pac Bell will simply set its tariffs to a reasonable
level (e.g. what what Bell Atlantic is already charging).


An additional piece of data that has recently come to light is that the
NSF subsidies to regional networks REQUIRE that the service provider
the regional buys service from be connected to all four NAPS! Why?
Wouldn't it be better to require "full connectivity" or even even trust
the regional to spend the money on a provider who can serve their
needs?  I suppose this is the only way NSF can force providers to
connect to all of the NAPS (except for MCI who NSF is PAYING to connect
to all the NAPs as part of the vBNS award).


        Wouldn't it be funny if NSF funded four NAPs and no one came?


I'll be they're thinking exactly that.


In summary, despite their (assumed) best intentions, NSF is grossly
confusing and distorting the existing commercial marketplace both by
forcing flawed technical interconnect architectures on commercial
providers and by providing direct financial support to the regional
networks who are directly competing with subsidized commercial
concerns.


NSF created the current Internet industry and they deserve its
gratitude, however, it's past the time when they should have declared
victory and let the child grow up on its own.