[2532] in Commercialization & Privatization of the Internet
Van Houweling's testimony
daemon@ATHENA.MIT.EDU (Eric.M.Aupperle@um.cc.umich.edu)
Fri Mar 13 22:17:47 1992
Date: Fri, 13 Mar 92 18:03:46 EST
From: Eric.M.Aupperle@um.cc.umich.edu
To: com-priv@psi.com
Hearing on the National Science Foundation Network
U. S. House of Representatives
Subcommittee on Science
March 12, 1992
Testimony of Douglas E. Van Houweling
Merit Network, Inc.
Mr. Chairman, members of the Committee, I am pleased to appear on
behalf of Merit Network, Incorporated, a not-for-profit organization
of nine Michigan universities. Those universities include: University
of Michigan, Saginaw Valley State University, Michigan State
University, Wayne State University, Eastern Michigan University,
Oakland University, Western Michigan University, Michigan
Technological University, and Central Michigan University.
As you know, Merit is responsible for the management of the
NSFNET backbone service under a cooperative agreement with the
National Science Foundation (NSF).
You have requested that I address the following in my
testimony:
-- The current arrangement for operation of NSFNET;
-- The NSF's plan for recompetition of the award for operation
of the NSFNET backbone; and
-- The key issues Congress needs to consider to help ensure a
successful evolution of the current Internet to the NREN.
Before I address these specific points, let me make a few
comments on the history behind the current cooperative agreement
between Merit and the NSF.
A Brief History
Merit has been involved in data communications networking
since the very beginning of the technology's development in the United
States. Established in 1966, Merit began operating an
inter-university packet switching data network in 1972. Today Merit
provides the State of Michigan with Michnet, a state-wide data
communications network. We are proud to have been pioneers in the
development of this critical technology for the future.
The current management and operation of the NSFNET backbone
service is based on Merit's cooperative agreement with the National
Science Foundation (NSF).
In 1985, the NSF established five national supercomputing
centers, and in 1986 linked those centers together at 56 kbps, thus
establishing the original NSFNET. That original network proved so
useful that it quickly became overloaded, and applications such as
remote computing were increasingly difficult to perform. In addition,
a number of regional networks wished to connect to the supercomputing
centers and to each other.
As a result, in 1987 NSF solicited ~ proposals from
organizations "to manage, operate, and continue development of a
national ... network." A key point in that solicitation was that "the
managing organization shall apply its expertise and creativity in
devising innovative approaches.... proposers are encouraged to suggest
alternate architectures and approaches that may be more appropriate,
more cost effective, or offer better service."
The solicitation also stated:
"The purpose of the NSFNET program is to provide scientists
and engineers with a national computer communications network that
will evolve to a national internetwork system for improving
communication, collaboration, and resource sharing.... NSFNET is
dynamic. It will change with evolving network affiliations, improved
technologies, competing communications costs, varying traffic load,
and other similar factors."
The solicitation went on to describe NSFNET's overall architecture to
be a "three-level hierarchy:
(1) A transcontinental 'backbone' network, interconnecting
(2) a number of autonomously administered 'second-level'
networks.... each of which interconnects
(3) as many as 30 academic, industrial, and/or government
research campus networks."
Merit believed that higher education participation would be
critical to the success and to the community's use of the backbone.
Further, Merit's analysis of the needed capabilities, its
understanding of the available funding from NSF, and the resulting
required level of effort led to the conclusion that successful service
would require a partnership with the computing industry and the
telecommunications industry.
On August 14, Merit submitted a proposal in partnership with
IBM and MCI. Two key aspects of that Merit proposal were:
1) The design and deployment of a 1.5 million-bit-per-second
(T-1) network from the beginning, with a option for providing 45
million-bit-per-second (T-3) service after 1990 if NSF so desired;
2) Significant cost sharing, including hardware and software
from IBM, connectivity from MCI, and funding from the State of
Michigan's Strategic Fund.
In total, this cost sharing was more than double the proposed
five-year NSF budget of $14 million.
It is important to underscore that the NSF process was a
highly competitive one. We understand that a number of companies in
the telecommunications and computing industries as well as other
universities submitted proposals.
On November 19, 1987, NSF announced a five-year cooperative
agreement with Merit for the design, engineering, construction, and
operation of a national backbone network service. NSF had the
capacity to extend and modify the agreement, develop additional
agreements and subcontracts, and provide additional funding as needed
over the life of the agreement.
~ Project Solicitation for Management and Operation of the NSFNET
Backbone Network, National Science Foundation, OMB 3145-0058.
In July 1988, only seven months after the award, Merit and its
partners deployed the backbone on schedule and within budget. Since
that initial deployment, the NSFNET services have enjoyed phenomenal
success due to NSF's leadership and the hard work and investment of a
number of industry participants. The traffic on the backbone has
grown almost 7,000 percent, an average of 11 percent compounded every
month, and new applications and uses are constantly emerging. This
has been a challenge to manage, but even more of a challenge has been
the growth in the number of networks that are now reachable via the
NSFNET service. This number has doubled every year and is now
approaching 5,000 networks worldwide.
In fact, the growth of the network and its use was so great
that it posed significant technical challenges. Recall that, at the
time, NSF and the Merit partnership were working on the leading edge
of a new internetworking technology. Some of the technical hurdles
were very difficult to overcome.
To accommodate this extraordinary growth, the backbone service
was continuously upgraded -- new connections were added, speeds
increased, and routing and network management technology enhanced.
Both NSF and the partners added new resources. NSF's additional
funding commitment was augmented by continued major cost sharing by
IBM and MCI.
The option to upgrade the network to T-3 speeds was exercised
by NSF for four reasons:
1) Many of the emerging applications, such as graphic user
interfaces that facilitated visualization of medical images and other
applications,~ such as ozone depletion, air pollution and fuel
combustion studies, as well as the emerging distributed file systems,
required broader bandwidth.
2) Many federal studies, including the September 1989 FCCSET
report as well as the OSTP program, proposed the need to upgrade the
speed of the network over time to T-3.
3) The projections of the extraordinary growth clearly
indicated that the T-1 network would be congested unless additional
capacity (e.g. by an upgrade to higher speeds) was deployed.
4) Finally, one of the overriding goals of all parties
involved in the NSFNET backbone service has been to continue to push
the limits of communications technology and maintain the U.S.
competitive advantage in this field.
Leading-edge technology has been required all along the way.
The challenge of incorporating such developmental technology in a
production network has presented all of us with many problems, but it
has also stimulated United States' research, industrial and commercial
leadership in fast-packet switching data networking.
Organizational Change
As the backbone network service grew in complexity, and was
re-engineered, the underlying organization also had to evolve.
Increased focus and resources were needed to keep pace with the
increasingly complex technical, business and policy environment.
~ Grand Challenges: 1993 High Performance Computing and
Communications, FY 1993 U.S. Research and Development Program,
Supplement to the President's FY 1993 Budget.
Therefore, with the agreement of NSF, on September 14, 1990,
Merit, IBM, and MCI announced the formation of a new, not-for-profit
corporation, Advanced Network & Services, Inc. (ANS).
The certificate of incorporation states that ANS:
1) "is ... dedicated to the advancement of education and
research in the interest of improving the ability of the United States
to compete in the global economic environment";
2) "will concentrate on computer networking and related
services,";
3) "shall help establish a high-speed computer network which
will be maintained at the leading edge of technology,";
4) "will ... help to expand the access to and interchange of
information technology resources among academic, government and
industry users,"; and
5) "will engage in research and development work which will
...contribute to United States preeminence in high speed
networking...."
ANS' board, of which I serve as Chairman, is broadly
representative, its members drawn from industry and higher education
across the nation. Both IBM and MCI made major contributions in the
form of grants to ANS' establishment. In addition, both IBM and MCI
bear substantial cost in providing equipment and services to ANS.
With the approval of the National Science Foundation, Merit
contracted with ANS for backbone network service. The goals were:
1) To provide a stable nationally representative
organizational and financial platform for the future of the NSFNET
backbone service and its successors.
2) To begin the process of privatizing the network.
3) To provide a foundation for interconnecting commercial
service providers with higher education and research enterprises.
It is important to note that there were no government asset
transferred to ANS at the time of its establishment. The network
hardware itself is owned by ANS, and the communications facilities are
leased by ANS from MCI and others. Federal agencies, regional and
state networks, other commercial networks, universities and private
industry pay for their attachments to the ANS network.
The formation of ANS brought to the network's development an
increased amount of private sector participation -- a goal which
Congress recognized in passing the High Performance Computing Act of
1991. The formation of ANS also was heralded by Senator Al Gore
(D-TN), who stated that "Just as private contractors helped build the
interstate highway system, this new corporation will help build the
national information superhighway today's information age demands."
This increased participation of the private sector, spurred by
ANS's entry into the field, also broadened the funding base and
increased competition in the network's development. For example, ANS
has also submitted proposals to provide services to emerging state
networks with other interexchange carriers. In addition, new test
equipment vendors have been integrated into the network to improve
network management for all users at very high speeds.
To accomplish the third goal, ANS on May 30, 1991 established
a for-profit subsidiary, ANS CO+RE Systems, Inc., to serve commercial
customers and link them to the research and education community.
Discussions with representatives of the regional networks, FARNET, and
the NSF were held to ensure that the cost of providing commercial
services are completely reimbursed through non-governmental sources.
Fees from commercial use of backbone services, minus operating
expenses and taxes, are returned to ANS for reinvestment in the
network infrastructure.
The creation of ANS CO+RE and its work with regional networks
blazed the trail for major segments of the Internet to carry
commercial traffic, creating many new commercial Internet service
providers and taking a major step forward in pursuing the vision of
the original drafters of the program.
The Results
Less than four years later, it is instructive to assess the
results of the NSF's visionary plan.
First, access to the network has exceeded the most optimistic
visions. The solicitation said, "It is anticipated that over the next
five years NSFNET will reach more than 10,000 mathematicians,
scientists, and engineers at 200 or more campuses and other research
centers."
In fact, the NSFNET today provides access to millions of
scholars and researchers in over 1000 institutions across the United
States. Over 650 colleges and universities are connected representing
approximately 80 percent of the nation's student population and 90
percent of the nation's federally sponsored research.
Further, NSFNET now provides access to more than a thousand
high schools and several hundred libraries through the joint efforts
of the NSF and the regional networks.
This data network is going to open up entire new vistas of
information and learning techniques for America's students from
kindergarten to post-graduate institutions. I believe it will spark a
revolutionary change in the way we prepare our children for the
challenges of the future.
This evolving network is clearly an example of where economies
of scale make a difference in serving many diverse communities.
Second, federal funds have been leveraged in an extraordinary
fashion. Every dollar spent by NSF on the backbone service is matched
by more than four dollars from Merit and its partners. Further, each
of these dollars has yielded an investment at least ten times larger
in regional and campus networks.
Mr. Chairman, the bottom line is, every Federal dollar has
stimulated at least a forty-dollar investment in an emerging
technology critical to the nation's international competitiveness.
With a small expenditure of federal funds, the government has spurred
the development of an entirely new set of technologies and
applications that will dramatically enhance our ability to compete in
world markets.
Third, a new industry has begun to develop in which the United
States not only leads technologically, but in the market. American
companies lead the world in internet technology, and are growing
rapidly. Not only are large communications carriers like Sprint,
AT&T, MCI, Ameritech, and Bell Atlantic working vigorously to
establish their presence, but smaller companies and regional and state
networking organizations are growing rapidly. Merit and ANS are
simply two of the many organizations in this rapidly growing
marketplace.
In keeping with the goals of the High Performance Computing
and Communications Program, the NSF program has been an extraordinary
success in stimulating the growth of a critical new technology. The
program grew from a federal government commitment to seed an infant
technology, was helped and nurtured by a committed partnership of
educational and research institutions, state government and private
enterprises, and it is now demostrating its potential. Already, the
United States is the world leader in exporting this new communications
technology and its many related applications. Those increased exports
are making a positive difference in our balance of payments.
This network is envied worldwide. It has allowed the U.S. to
lead in building a high-speed data communications infrastructure. We
are pioneering technological applications that will be among the most
critical for research, education, and business in the 21st Century.
Fourth, and perhaps most important in the long term, is that
government, higher education, and industry are working cooperatively
to build a critical infrastructure for our nation. No one argues with
the proposition that knowledge and information are critical
commodities for our nation's future.
Key Issues For The Future
The NSFNET example and experience is providing the United
States with the tool it needs to succeed in that future. NSF's vision
and program have led us to this uniquely American approach to
innovation. With a modest investment, they have energized a nation.
Their program plan and strategy for introducing future technology will
sustain the momentum. We believe NSF's program plan for recompetition
of the cooperative agreement is appropriate, but entails substantial
technical and management challenges.
Congress should consider the following key issues to help
ensure a successful evolution to the NREN.
1) The successful triad of government, academic, and private
industry should continue because it spawns innovation and technology
transfer which makes this country stronger.
2) Any future program should seek to stimulate investment from
private industry as the current program has so successfully done.
(3) Congress should create an environment in which the natural
tendencies for fragmentation of federal efforts are overcome by a
coordinated program led by the NSF.
Merit and its partners have been pleased to have played their
part in this extraordinary national program.
