COOK Report Predicts Winners of NSFnet Solicitation

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

Posted-Date: Wed, 9 Feb 1994 16:13:45 -0500
Date: Wed, 9 Feb 94 15:51:14 EST
From: cook () pandora sf ca us


In view of the meeting of the National Science Board that begins tomorrow, it
seems appropriate to share with com-priv what the COOK Report sent to
subscribers on January 30.  Our article follows:
_____________


MCI AND ANS WILL WIN VBNS


DECISIONS HAVE BEEN MADE BY NSFNET DIRECTOR WOLFF -
WE PREDICT WHO WINS VARIOUS COMPONENTS OF NSF 93-52


NSF SPREADS THE PRIZES WIDELY BUT
INTER-REGIONAL CONNECTIVITY COMPONENT LOOKS LIKE CAN OF WORMS


We believe that we have been able to piece together with a great deal of
certainty the identities of the winners of at least the first two components of
NSF Solicitation 93-52.  Of course the awards won't become official until the
National Science Board ratifies Steve Wolff's decision at its February 10 - 11,
1994 meeting.


Herewith the prize winners:


1.  The very high speed backbone:  MCI with ANS as subcontractor.


2.  The Network Access Points:
        New York City - Sprint
        Chicago - Ameritech and Bellcore
        Washington DC - Metropolitan Fiber Systems
        California - BARRnet & PacBell


3.  Inter-regional Connectivity


We hear from one very well placed source that CoREN will be the recipient of the
inter-regional connectivity award.  The question of greatest interest is who the
transport provider for CoREN will be.  We strongly suspect it will be MCI
because MCI has been the CoREN partner from the inception.  (We wrote about
CoREN as long ago as May of 1992.)


While we have heard that MCI has failed to give CoREN a price on ATM
connectivity, we believe that the most rational scenario is for MCI to emerge as
the CoREN service provider.  When one stops and thinks about the understandable
concerns of the NSF that there be a stable transition for the network the
advantage of having MCI as the transport level provider of inter-regional
connectivity for the largest mid-levels are obvious..


Under this scenario, while it will undergo a metamorphosis, the current ANS
backbone could survive.  On February 3 ANS is expected to move its network
operations center into new Ann Arbor headquarters provided by Northern Telecom.
The core nodes in the network collocated in MCI POPs on its national backbone
will not have to change.  Conversations inside ANS reveal that most ANS NOC
staffers believe they will soon be offered jobs by MCI.  If MCI and CoREN stay
together, the mid-levels will have used this part of the solicitation to execute
a take over of the ANS backbone and squeeze ANS out of the picture just as they
believe that ANS intended to squeeze them out earlier.


While that would seem to be the most rational scenario, it is likely too much to
expect rationality given the checkered history of NSF 93-52 We have heard an
opposing view that suggests the current ANS backbone is unlikely to become
CoREN.  Since CoREN is to be ATM based the routers will therefore have to be
different.  It also emphasizes a recent falling out between CoREN and MCI
apparently based on MCI's alleged unwillingness to give credible ATM delivery
dates and pricing to CoREN.  This fact and the fact that only about half the
major mid-levels joined CoREN will ensure that there will be a lot of
disruption.  When you combine CoREN's insistence on starting with ATM with the
fact that most of the rest of the mid-levels are being funded to "do their own
thing," a smooth transition is unlikely.


We think this latter scenario is the more credible and, if accurate, could well
lead to a third extension for the current incumbents.


4.  The routing authority may go to MERIT which we know did put in a bid for it.
If it did, all parties would maintain the convenience of continued collocation
in Ann Arbor.  We emphasize that this is only as guess since we concentrated our
efforts in learning about the other parts of the proposal.


MCI as an Astute
Strategist


Lets go back to the vBNS.  While IBM thought it was playing for big stakes in
the original Merit Cooperative Agreement, MCI emerges as the hands down winner.
MCI kept well in the shadows during all the controversies that surrounded ANS.
It now stands ready to pick up the pieces.  As prime bidder on the vBNS it is in
a position to call the shots.  Vint Cerf who is one of the most respected and
trusted senior founders of the Internet has returned to the MCI fold.  Vint will
quite clearly be in charge of making things work.  In contrast to Al Weis, Cerf
will bring his technical expertise in the network technology and his familiarity
with the internet culture.  All these are characteristics that will make the NSF
happy and help take the heat off of an otherwise controversial solicitation.
The new arrangement will very likely shunt ANS off into the role with NT as a
developer of routing and switching capability.  With MCI in the driverUs seat we
would also hope that other switching hardware would also be used.


The players now seem so sure of the direction of the awards that they are
changing employers.  In addition to Cerf moving from CNRI to MCI in early
January, on January 18 ANS lost another vice president.  Phill Gross, the Chair
of IETF who had worked for Cerf at CNRI before going to ANS in early 1991 moved
back to MCI where he will again work on network architecture under Cerf.  (We
NOW have been told that Gross will not report to Cerf.) We are told that Gross
took most of his staff with him to MCI.  One way to solve the problem of ANS as
"enfant terrible" of the Internet is to relegate it to a small and harmless
corner of the sandbox.


What About Sprint and AT&T?


We have been told as early as last September that AT&T bid zero dollars for
providing an OC3 vBNS.  We have heard this since then from five different
people.  Why did AT&T not win?  We have been told that its proposal lacked
technical credibility.  Perhaps.  We are not equipped to judge.  Of course
nothing helps add to credibility like being in charge of the present service.
Of course we have also been told that MCI added an option to expand to OC12
before the end of the cooperative agreement.  Nothing like the advantages of
being there first!


We are less certain why Sprint lost out.  Perhaps DOE's continued desire to
award them the ESnet contract weighed against them?  Again they were not in that
critical driver's seat.  Among those in the know the view is that they were
awarded the New York City NAP as a consolation prize.


The NSF seems to have taken care to spread the spoils as widely as possible.
Thus Bellcore and Ameritech representing the RBOCs get the Chicago NAP.  While
we hear from two sources that BARRnet gets the California NAP, one other source
whose judgment we respect refutes this claim.  A fourth source stated it would
go to a combination of BARRnet and PacBell.


How will all this shake out?  While it is possible to imagine a worse outcome,
we will reserve judgment until the details are known.  Certainly there is as
much "pork" here as there is technology development.  An OC3 ATM backbone for
the supercomputer centers could be designed in such away as either to minimally
advance or radically advance the state of the art.  With Vint Cerf's reputation
riding on this line we will hope for the best.  Of course why we need to install
real time capability for a Cray in California to crunch the same problem as a
Cray in New York is much less clear.  Though costly these machines are much more
common than they were a few years ago.  In the opinion of some it is also much
more cost effective to link them on the same LAN or MAN.


_______________________________________________________________
Gordon Cook, Editor Publisher:  COOK Report on Internet -> NREN
431 Greenway Ave, Ewing, NJ 08618
cook () path net                                   (609) 882-2572
Subscriptions: $85 individual; $175 non profit; $500 corporate site license
_______________________________________________________________