[1613] in Commercialization & Privatization of the Internet
Vint Cerf article
daemon@ATHENA.MIT.EDU (Telecom-Editor)
Wed Dec 4 18:10:22 1991
Date: Wed, 4 Dec 91 21:57 GMT
From: Telecom-Editor <0003101692@mcimail.com>
To: "com-priv@psi.com" <com-priv@psi.com>
ANOTHER READING OF THE NREN LEGISLATION
Vinton G. Cerf
Editor's note: The somewhat controversial views espoused by Jay
Habegger in his article are by no means universally held. The
following article, in point-counterpoint fashion, provides an
entirely different perspective on the impact of the NREN legislation
on telecommunications policy. The author, Vint Cerf, is widely known
throughout the industry as a result of his involvement in the
original ARPANET, which has now evolved to the Internet. Cerf is
also chairman of the Internet Activities Board and considered one of
the world's foremost experts on computer networking and the
Internet. -- TV
When considering the impact of the National Research and Education
Network (NREN) legislation, it is important to realize that the
House and Senate bills do not mandate that the NREN be built by
government. It mandates that an NREN be built -- but leaves open the
method and means. This makes the enabling legislation ambiguous, but
that is actually preferable to having Congress design the system!
The Internet was originally a 1970s experimental research project
sponsored by the Defense Advanced Research Projects Agency (DARPA).
It centered on the ARPANET, but linked hundreds of local area
networks and several other wide area or regional area nets
(satellite and mobile packet radio). In the early 1980s, a
consortium of universities, not directly sponsored by DARPA,
approached the National Science Foundation for support of a system
they called CSNET, which would use public X.25 technology and dialup
technology to link to the Internet using the TCP/IP protocols or
higher level electronic mail services. CSNET rapidly became
self-sustaining (a condition of NSF support). In the mid-1980s, NSF
embarked on a supercomputing center support effort and associated
with that a network for linking them together. The NSFNET was a
cooperative effort among NSF, MERIT, IBM, MCI, and the state of
Michigan. Although it began with a low-speed, 56-kbps backbone (same
as ARPANET), it quickly moved to 1.5 Mbps, using switches developed
by IBM and intercity links provided by MCI. In the early 1990s, it
has achieved a 45-Mbps backbone capability using a new set of IBM
switches that cover about half the nodes of the 1.5-Mbps backbone
and that will eventually cover the entire NSFNET backbone.
It is critical to understand that except for the local area networks
on campuses or industry sites, the US portion of the Internet has
always been based on transmission facilities provided by commercial
telecommunications carriers. Because the carriers typically have not
owned the switches, however, the entire complex of networks has been
operated as a private network (in the same sense that a corporate
net is considered private if the corporation owns or leases the
switches). In the recent past, not only have private sector,
commercial service providers emerged (e.g., Performance Systems
International, CERFnet, UUNET/Alternet, Infonet, and ANS/CO+RE), but
traditional common carriers, such as Sprint, MCI, and AT&T, as well
as the RBOCS, have begun various trials of metropolitan area or wide
area packet-switching technology capable of supporting TCP/IP or OSI
protocols.
Considered internationally, the Internet is not a hierarchical
network so much as it is a mesh of backbone facilities that link
regional and local systems together. The NREN effort in the US is
likely to change the face of the Internet in the US, but it is not
likely to change the general international architecture of the
Internet. Taken in its international context, the Internet is just
as much an amalgam of private, commercial, and government network
facilities as found in the US. There are over 5000 networks in the
Internet linking over 570,000 computers and on the order of 3
million users.
GIGABIT NETWORKING
With respect to gigabit networking, sheer economics makes it
unlikely that a gigabit national network could be funded solely by
the US government. The NREN enabling legislation is a drop in the
bucket by comparison with the expenditures required to create the
existing, ubiquitous telephone network and its already gigabit
backbone. A gigabit national data network only makes sense if it is
derived from the ongoing investment by the telecommunications
carriers in optical-fiber transmission facilities across the US.
What the NREN legislation can do is focus on applications that might
not have been considered by the telecommunications carriers that are
of particular interest to the research and education community. No
one really knows which applications really demand end-to-end gigabit
bandwidth. The DARPA and NSF sponsored research on gigabit
networking is partly aimed at finding out by experimentation in a
series of five testbeds.
Many telecommunications carriers are participating in the testbed
program, along with government laboratories, private industry, and
university researchers. The effort could not have begun without this
collaborative participation, especially with industry and the
carriers. The carriers have provided gigabit fiber and SONET
transmission technology. Industry has provided switches and special
interfaces allowing computers to transmit and receive at gigabit
rates. The universities and government labs are designing and
building high-speed applications often involving supercomputers and
high-power workstations. Various switching architectures are being
tested experimentally in this effort, and the existing Internet
protocols (e.g., TCP/IP and OSI) will be stressed to run at these
speeds. It is possible that new protocols will be required to
exploit fully the facilities inherent in asynchronous transmission
mode switching and SONET transmission.
Eventually, it is anticipated that the gigabit transmission and
switching technology emerging from the gigabit testbeds will be
incorporated into the US portion of the Internet, forming the
beginning of the NREN. Expansion of this US part of the Internet is
expected in two dimensions. The enabling legislation supports
subsidy for a wider population of user sites in research and
education, while also encouraging much higher access and backbone
speeds. As always, there will be some sites that are prepared to
apply the highest possible speeds supported in the system and others
that can function adequately with lower speed access. This is as
true in the present telephone network today as it is likely to be in
the future national network.
DIVERSITY IN THE INTERNET
The present US Internet is made up of many different parts. The
federally subsidized components, such as the NSFNET, NASA Science
Internet (NSINET), Energy Sciences Net (ESNET) and DARPA Test Net
(DARTNET), have agreed to interwork and to carry each other's
traffic.
The commercial networks (PSINET, CERFNET, UUNET/ALTERNET) are linked
in a commercial internet exchange (CIX) and are, in turn, linked to
the NSFNET backbone. Most of the midlevel networks are linked to
NSFNET and/or the commercial networks. International connections are
brought into the US through government agreements or as a result of
business negotiations by the commercial nets. Although this is not a
perfect process, it has served as a framework to combine the
interests and contributions of thousands of network providers,
including college campus and corporate local nets.
READING THE NREN LEGISLATION
The NREN legislation can be read to mandate a centrally managed, and
even centrally owned, system but this is not a required reading and,
perhaps, even a fatal misreading of the vision. Achieving a
ubiquitous, affordable, and high-speed networking capability for the
research and education community will require weaving together a
remarkable amalgam of interests, including most especially the
private sector. The US government has played a key role in the
creation of the technology and the stimulation of interest in
service provision in the private sector. It seems that this
important initiative, a first step towards the creation of a
national, high-speed information highway system, can only be
achieved through a blending of interests and investments. The
telecommunications industry has a key role to play if ubiquity and
industry access is to be realized.
The US government can contribute subsidizing funds to create
markets; can foster development of new technologies, applications,
and standards; can set up frameworks that encourage private sector
investment including, for example, allowing commercial service
providers to link to the NREN to serve the NREN user community; and
can foster development of freely available software that can be
acquired and shaped by industry into commercial products. It has
done all of these things in connection with the existing Internet
system and has spawned new companies, products, and services meeting
the needs of the growing Internet population. It can achieve similar
successes through the proposed NREN effort.
Vinton G. Cerf is vice president of the Corporation for National
Research Initiatives (CNRI) and chairman of the Internet Activities
Board. He is also a member of Telecommunications magazine's
editorial advisory board. His opinions express personal views and do
not represent positions of his employer or of the Internet
Activities Board.
***
This article first appeared in the November 1991 issue of
Telecommunications
Copyright 1991
Telecommunications magazine
