Jacking in from "The Envelope Please" port: From Brock N. Meeks

[David Farber <farber () central cis upenn edu>]

CyberWire Dispatch // Copyright (c) 1994 //


Jacking in from "The Envelope Please" port:




Washington, DC -- After more than a year of poker-faced silence, MCI has
finally taken the wraps off its long-awaited Asynchronous Transfer Mode
(ATM) strategy.  Today, MCI is smiling.  Hell, they're probably laughing.


Enduring any number of criticisms, from both the press and the industry,
MCI has quietly been working since July of 1993, building and testing a six
node ATM test network.  The network is built around Northern Telecom's (NT)
Magellan Gateway ATM switch.


The state of the network has sufficiently progressed that MCI has announced
it's now conducting customer trials on nationwide basis that includes
applications ranging from desktop videoconferencing for brokerage offices
of Bear Sterns to interactive medical imaging for doctors to remote
visualization for scientists.  "These real-world ATM customer trials will
enable us to better gauge the possibilities and challenges of ATM prior to
finalizing our commercial offerings," said Paul Weichselbaum, vice
president of data marketing for MCI's Business Markets group.


The controversy surrounding MCI's thought-to-be nonexistent ATM technology
has at times been withering.  When the National Science Foundation (NSF)
recently announced MCI as the winning bidder to build and run its
next-generation high speed computer network, known as the Very High Speed
Backbone Service (vBNS), industry pundits were rocked on their heels.  How,
they wondered, could the NSF have possibly awarded the prestigious vBNS
contract to a company with no professed ATM strategy?


Charges and counter charges raged across the Internet.  Sprint, a losing
bidder for the vBNS, filed a protest, claiming in part that the government
through the NSF award to MCI, would actually be funding that company's
research and development of ATM technology.  That process sufficiently
smacked of industrial policy, a means whereby the government purposefully
chooses one private sector firm over another, giving the winning firm a
competitive advantage on the backs of taxpayers.


The hue and cry of the Net community didn't escape congress.  During the
last week of April, Rep. John Conyers (D-Mich.), who heads up the House
Government Operations Committee, sent a sharply worded letter to NSF
Director Neal Lane, warning that he'd received allegations that NSF's award
to MCI was conducted through a "sham competition."  Conyers also echoed
Sprint's charge that the government might be funding development MCI's ATM
technology.


MCI's now enjoying a belly laugh.


Slight of Hand
==============


But the story's not as cut and dried as it might seem.  MCI still refuses
to divulge exactly what technology and in what configuration it actually
proposed in its winning bid.  It's safe now to assume, that the cornerstone
of MCI's vBNS proposal was its NT-driven ATM network.


MCI claims its ATM network has been in place since July of 1993.  At first
blush, that seems impossible.  NT didn't even announce its Magellan Gateway
ATM product until a month later, in August of 1993.


"We gave MCI some first cut, prototype ATM switches in July," said an NT
technology wonk.  NT turned up those switches later that same month, and
the service bucked and spit all the usual technological bugs.  But it was
in place.  "A lot of software changes have been done since then," the NT
wonk told Dispatch.  "This was a learning experience for all of us," he
said.


NT refused to say whether or not they had given the technology to MCI.
"Nothing is free in this world," the NT wonk said.  "But we partnered with
MCI to a large extent to get the benefit of the experience."  NT made its
last software upgrade to the switches in February he said.


Commercial Use Approved
=======================


The ATM announcement by MCI also helps tightens up a rather disturbing
situation that has been unraveling in the last few months.  That situation
involves a statement made by NSF's Director of Networking Computing Steve
Wolff regarding pre-approval of the vBNS provider to also sell commercial
services across that network.  The thought that the government would be
paying $50 million to some company to provide a high speed computer network
for the nation's supercomputer centers and then turn around and allow it to
also make money from a supposed government funded network was, well,
unpalatable.


It now appears that Wolff may have had little choice in the matter.
There's not a chance in hell that MCI would have agreed to use its ATM
network solely as a long haul megabit relay service for the pittance (by
eventual commercial market standards) NSF was providing under the vBNS
contract.  To agree to provide ATM service to only NSF via the vBNS deal
would have been commercial suicide.  So MCI cut a deal with the NSF:  Let
us also use the ATM network for commercial purposes or it's no dice.


Strong arm tactics?  Maybe.  But Sprint would have been in the same
position.  It has only one ATM network and it's being used for commercial
purposes, too.  And by comparison, the nation's other as-yet-to-be realized
high speed computer link, the Energy Department's ESNet -- a contract
eventually worth $55 million -- was *required* to be provided by a company
that also used it for commercial purposes.  The rub there being, no company
was going to let its commercial product tank, thereby stiffing the
government users.


But this damn vBNS project is like an onion:  As soon as you peel one
layer, there's another.  In this case, the question that remains is:  Why
has the NSF chosen to award the vBNS under the loose structure of a
cooperative agreement?  Under this arrangement, the NSF grantee isn't held
to strict, binding contractual terms.  If MCI doesn't deliver on promised
services to NSF under the vBNS agreement, it can still be paid, and
according the NSF's experience with the ill-fated NSFNet, a failed service
provider still gets a check in the mail.


Conyers asked similar questions in his letter to NSF Dir. Lane.  If MCI has
a soon-to-be commercial ATM network, why doesn't the NSF simply contract
for those services and put some real enforcement teeth into the deal?  The
NSF will answer, no doubt, by claiming that as a research and education
project, the technology will be "bleeding edge" and by its very nature,
unstable and often quirky.


Even MCI admits that its current ATM network isn't bulletproof:  "We are
continuing to test ATM technology with our customers while allowing time
for the standards, equipment and market to mature," MCI's Weichselbaum
said.  "Our approach is to then harness this technology to offer meaningful
and differentiated services to our customers," he said.


In the meantime, MCI will be "experimenting" with our nation's most
prestigious R&D facilities, that's all.  I know I'll sleep better now.


So, NSF will tell you (and has probably already told Conyers), that when
you're pushing the technological envelope, the service provider shouldn't
be held to the strict confines of a contract. After all, building the
nation's premiere high speed network isn't exactly like the Pentagon's
demanding procurement  specifications for claw hammers.


Meeks out...