IMPORTANT INFO FROM NSF ON FUTURE OF INTERNET

BIOSCI Administrator biosci-help at net.bio.net
Mon May 1 21:21:49 EST 1995


I just pulled this over from NSF and thought that it would be of
interest to our readers in the U.S.

				Sincerely,

				Dave Kristofferson
				BIOSCI/bionet Manager

				biosci-help at net.bio.net


From: STISSERVE - NSF E-mail Server <stisserve at nsf.gov>
Subject: nsfmci.txt (complete) ascii


------ begin of nsfmci.txt -- ascii -- complete ------
Title  : NSF-MCI - BACKGROUND INFORMATION ON THE INTERNET/NSFNET
Type   : News
NSF Org: OD / LPA
Date   : April 26, 1995
File   : nsfmci


For More Information contact:
    Beth Gaston, (703) 306-1070
    e-mail: egaston at nsf.gov


                         FREQUENTLY ASKED QUESTIONS


With the new architecture, will I have to pay more for my
Internet connection?

     Most likely, no.  Institutions most affected are the
approximately 2,000 universities, research organizations and
governmental agencies who have been the primary beneficiaries
of NSF provision of the NSFNET.  These institutions, which
currently pay between $10,000 and $60,000 for links to the
Internet, will see an average increase of $1,500 per year,
which is being provided to their Regional Network Providers by
NSF on a declining scale: 100 percent the first year; 75 the
second; 50 the third; 25 the fourth; and zero the fifth.
However, many people believe that fierce private competition
and the economies of scale will drive the price for Internet
hookups down.

Will the new architecture affect how I use the Internet?

     No.  In the past, traffic between individuals traveled
over a combination of government and commercial networks --
the only change is that the NSFNET is no longer one of those
networks as it has been replaced by commercial networks.

Can I use the vBNS instead of a regular network?

    No.  The vBNS is a research network available only for
meritorious high bandwidth activities.  It will not carry
regular, commodity-level Internet traffic.

What is the Internet?

     The Internet is a collection of more than 46,000
networks, public and private, worldwide that have agreed to
use common protocols and exchange traffic.

Who is in charge of the Internet?

     There is no single person or agency in charge of the
Internet -- it is unofficially coordinated by the Internet
Society (ISOC), a voluntary membership organization whose
purpose is to promote global information exchange through
Internet technology.  A group of volunteers makes up the
Internet Architecture Board, which meets regularly to approve
standards and allocate resources.  The Internet Engineering
Task Force (IETF) consists of volunteers and determines
solutions to nearterm technical problems of the Internet.

What will the NSF's involvement in the Internet be?

     The NSF will continue to support the InterNIC, which
provides limited information services for the research and
education community, and directory and database and
registration services to the non-military Internet community.
The NSF will also continue to support the connections program
which links up to 200 new institutions to the Internet each
year; existing International Connections Management Services;
and CNIDR, the Clearinghouse for Network Information Discovery
and Retrieval, which looks for, promotes the compatibility of
and, occasionally develops software tools for the discovery
and retrieval of information from the distributed databases on
the Internet.


       BACKGROUND INFORMATION ON THE INTERNET/NSFNET



HISTORY

     The concept of the Internet is extremely confusing to many
because it is not a single entity with a single person or
company incharge.  Instead it is more than 46,000 networks,
both public and private, world wide that have agreed to use
common protocols and exchange traffic.

     The precursor to the Internet began about 20 years ago
with the U.S. Defense Department's ARPAnet.  The network was
designed to operate in the event of partial outages -- so if a
nuclear blast obliterated part of the network, the rest of the
system would still work.

     In the mid 1980s, as a part of its effort to provide the
US scientific research community with access to supercomputers,
the National Science Foundation provided funding to establish
several national supercomputer centers.  To allow researchers
to use the computer centers, NSF had to either physically
transport the scientists to the centers or give researchers a
way to access the computers remotely from their home
institutions.  In response, the NSF determined to use both
methods.

     Because the supercomputer centers are broadly shared by
the research community, with particular projects changing with
each funding cycle, a dedicated network to a small number of
specific sites was not a viable solution -- broad access is
necessary.

   In 1985, the NSF created the NSFNET, which is actually a
hierarchical series of networks for research and education
traffic within the US.  The NSFNET was based on ARPAnet
protocols.  The "crown jewel" of the NSFNET was a national
backbone service which provided free carriage for any US
research and education traffic that could reach it.  Additional
components were supporting the development of regional networks
to bring electronic traffic from individual institutions to the
national backbone service as well as a connections program to
help individual institutions connect to the regional networks.

     The national backbone service was to be used both as an
infrastructural network for the research and education
community and also as a testbed for network-related research.

     A great intellectual leap occurred when people realized
the network could be used for more than simply the transfer of
data between computers.  The popularity of data networking grew
wildly and the NSFNET became an essential infrastructure for
the US research and education community.  Commercial firms,
noting the popularity and effectiveness of this technology,
built their own national networks and firms like PSI, Alternet
and Sprint began providing network services like those
available on the NSFNET. To the extent that customers of
commercial firms were a part of or operated in support of the
research and education community, traffic was exchanged between
the commercial networks and the NSFNET.

     But the NSF's role is changing.  The NSF is withdrawing
>From providing commodity-level network service (since these
services can now be competitively obtained in the marketplace)
and has awarded a cooperative agreement for a very high speed
network service (vBNS) to serve as a research network.  The
vBNS will connect the supercomputer centers and other
competitively selected high-bandwidth applications to push the
limits of networking technology.

     The government no longer needs to support the
infrastructure of the Internet -- identical capability is
currently available from private firms.

     The "new architecture of the NSFNET" will consist of the
regional network providers (who have historically provided
access for individual campuses to the NSF-provided national
service) obtaining their interregional connectivity from
competitively selected Network Service Providers (NSPs).  NSPs
must agree: 1-to connect to hubs called network access points,
where the NSPs will exchange traffic; 2-to route and carry all
traffic to or from any research and education location; and 3-
to make the routes to a routing arbiter.  The routing arbiter
services will provide an inventory of all available routes for
electronic "traffic" and will make them available to network
service providers. -----------------
REGIONAL NETWORK PROVIDERS

Approximately twenty cooperative agreements for regional
network providers (RNP) have been awarded.  Under these
agreements, the providers are initially receiving funds
approximating any increase in costs caused by having to obtain
their interregional connectivity from a commercial Network
Service Provider rather than using the NSF-provided network.
The amount received by each RNP will decrease with regionals
expected to cover the full cost of their commercial
connectivity beyond the fourth year.

NETWORK SERVICE PROVIDERS

Network Service Providers such as Sprintlink, MCI and ANS
provide connectivity on a national basis.  Regional networks
connect to NSPs for interregional (national) connectivity and
connection to interexchange points called Network Access Points
(NAPs).  NSPs connect these NAPs and agree to carry all traffic
to or from research and education locations.

NETWORK ACCESS POINTS

Four Network Access Points (central hubs for regional networks)
have been awarded:

·     a New York NAP to Sprint
·     a San Francisco NAP to Bellcore with Pacific Bell as the
operator
·     a Chicago NAP to Bellcore with Ameritech as the operator
·     a Washington DC NAP to Metropolitan Fiber systems Inc.

The organizations establishing Network Access Points will each
receive about $800,000 from NSF over a five year period.  (This
funding is for research related to their NAP activity.)  The
organizations will capitalize the NAPs themselves and
anticipate they will profit by seeking customers for their
interconnectivity services.

ROUTING ARBITERS

     Two five-year collaborative cooperative agreements have
been made to offer routing arbiter services.  The routing
arbiter offers routing and other services to facilitate logical
interconnection of attached networks.  Routing Arbiter
activities include: promoting Internet routing and stability,
establishing the network framework and policy databases,
developing procedures to resolve problems between network
entities, developing advanced routing technologies, providing
simplified routing strategies and promoting distributed
operation and management of the Internet.

     Merit Network, Inc. and the University of Southern
California's Information Sciences Institute have each received
5year cooperative agreements for approximately $10 million each
to collaborate and provide routing arbiter services.

  At midnight on April 30, 1995, the NSFNET should no longer
be needed and the National Backbone Service which has been
provided by NSF will be turned off.  (It will not be turned off
unless all necessary new infrastructure is operating
successfully to ensure the uninterrupted availability of
services to the research and education community.)

  Most Internet users will see no change as the result of the
turnover. Institutions most likely affected by the turn over
are the 2,000 universities, research organizations and
governmental agencies who have been the primary beneficiaries
of NSF provision of the NSFNET.

     Interregional connectivity, formerly provided by the
NSFNET, will be replaced by that obtained by regional network
providers from commercial network service providers.  Any
increase in the cost of the services (as a result of the
regional providers now having to pay for what was formerly
provided by the NSF network) will be subsidized on a declining
scale: 100 percent the first year; 75 percent the second; 50
percent the third; 25 percent the fourth; and zero the fifth.

     Based on the figures received from the regional network
providers, it appears that the total increase in costs carried
by those networks as a result of the changed architecture will
be $23 million per year, or an average of $1,500 per year per
institution.  The average institution currently pays between
$10,000 and $60,000 per year for connectivity and related
services.

     However, many believe that the fierce private competition
will drive down costs even as the demand for bandwidth
increases.

vBNS CONNECTING SUPERCOMPUTER CENTERS

  MCI has been selected for a cooperative agreement with the
NSF to provide a very high speed experimental network service
(vBNS) connecting the NSF-funded supercomputer centers, and
allowing for research on high-speed networking.  This network
is not an operating infrastructure; regular Internet traffic
will not travel over this virtual network.  Research using the
vBNS may later be developed into technology for use by the
various commercial service providers.

CONTINUED SUPPORT

   The NSF will continue to support InterNIC, which provides
limited information services for the research and education
community, and directory and database and registration
available to the non-military Internet community.  InterNIC was
established in December of 1992

The NSF is also continuing to support:

·     The connections program which links up to 200 new
  institutions to the Internet each year.
·     Existing International Connections Management Services,
provided by U.S. Sprint Corp., in effect through the spring of
1996.
·     CNIDR, the Clearinghouse for Network Information Discovery
  and Retrieval will continue to research and promote the
  compatibility of software tools for the use of the Internet.

CONCLUSION

     While developing the NSFNET and guiding its early growth,
the NSFNET Program was concerned primarily with provision of
the physical infrastructure necessary to allow the US research
and education community to use data communication and
networking technology as a working tool.  Now that commercial
Internet services have grown sufficiently to provide the
necessary infrastructure to allow this basic physical
infrastructure to be purchased in the market place, the NSFNET
Program is refocusing its efforts on high levels of
connectivity, function and use for the research and education
community.

                         A GLOSSARY OF NSFNET TERMS




backbone - a top-level network, one that provides continental-
scale interconnectivity to several regional networks and/or other
backbones.

bit - acronym for binary digit

byte - a group of adjacent binary digits seen by the computer as
a "word." (A byte usually contains either seven or eight bits.)

broadband - a multi-frequency transmission channel.

CNIDR - Clearinghouse for Network Information Discovery and
Retrieval--NSF supported program to research and promote the
compatibility of software tools for the use of the Internet.

gigabit - one billion bits of information. Research testbeds,
including the new vBNS, are currently striving to develop
effective networks that can transmit a gigabit of data per
second.

High Performance Computing and Communications (HPCC) initiative -
a program to foster the full range of advanced computing
technologies, including existing supercomputer systems, and
special-purpose and experimental systems. It includes, among
other elements, high-speed networks that transmit billions of
data bits per second, large-scale parallel processing to solve
immensely complex problems by approaching them on several tracks
simultaneously, advanced algorithms and visualization displays to
depict phenomena ranging from the molecular to the
meteorological.

Internet - a worldwide collection of more than 46,000
interconnected computer networks. In the U.S., it consists of NSF-
sponsored networks that link thousands of university, government
and industrial research sites; networks sponsored by the defense
department's Advance Research projects Agency (ARPA), NASA and
the Department of Energy; and numerous privately operated
networks serving primarily individuals and businesses. Based on
the defense
department's now-decommissioned ARPAnet, its use has expanded
rapidly in recent years as more people use it for sharing
resources and information through remote logins, file transfer,
electronic mail and interest-group mailing lists.

InterNIC - Internet Network Information Center; provides limited
information services for the research and education community,
and directory and database and registration to the non-military
Internet community.

LAN - Local Area Network, a network that connects multiple
computers at a single site.

megabit - one million bits of information.

Metacenter - The organization of  computer centers into a
proactive, coordinated federation of centers, with greater
collaboration and less duplication.

national information infrastructure (NII) - Yet to be deployed,
it is a notion of an infrastructure consisting of a network of
communications systems and computer systems and trained personnel
supporting the exchange of information among all citizens and
residents of the US.

National Research and Education Network (NREN) program - a
component of the High-Performance Computing and communications
program that supports the development of high-speed research
network and support programs to provide a high-speed (gigabit-per-
second) research network with distributed computing capability.
It will link government, industry and higher education
communities. It will serve as a prototype for future commercial
networks. It will consist of two complementary components: the
Interagency Internet, an evolving, operating network which will
enhance existing federally funded research networks as it serves
as a medium for the deployment of advanced technologies in the
near future; and the Gigabit Research and Development effort for
broadband networks.

NSF Supercomputer Centers - Five centers partially funded by NSF
which provide supercomputer resources to researchers across a
range of disciplines, for both state-of-the-art research in
computer science and for complex scientific and engineering
problems which require their level of computing power. Four
centers were established in the mid-1980's: San Diego
Supercomputer Center and the University of California, San Diego;
Cornell Theory Center at Cornell University, Ithaca, NY; National
Center for Supercomputing Applications at University of Illinois
at Urbana-Champaign; and Pittsburgh Supercomputing Center at
Carnegie Mellon University and the University of Pittsburgh. The
National Center for Atmospheric Research in boulder, Colo., was
established in 1960 and later upgraded with supercomputer
equipment.

network access point (NAP) - Part of the new architecture for the
NSFNET-- a high speed network or switch to which a number of
networks can be connected via routers for the purpose of traffic
exchange and interoperation with the Internet. NAPs will serve as
hubs connecting network service providers and regional network
providers.

network service provider - Part of the new architecture for the
NSFNET -- network service providers offer connection on a
national scale. Regional networks hook into network service
providers for inter-regional (national) connection and for
connection to Network Access Points. NSPs connect the network
access points and agree to carry all traffic to or from research
and education locations.

NSF - National Science Foundation, an independent agency of the
federal government that provides support for research and
education in the sciences and engineering.

NSFNET backbone service - With the new architecture, the NSFNET
backbone no longer exists. It has been turned over to a variety
of commercial concerns, with government support decreasing on a
sliding scale over five years. Now, regional network providers
obtain interregional connectivity from Network Service Providers
with traffic exchanged at Network Access Points and governed by
Routing Arbiters.

NSFNET program - A computer networking program established by NSF
chiefly to allow exchange of information and access to remote
resources within the research and education community. The NSF
continues to support the connections program, promoting research
and educational institutions to hook into the Internet;
international connections management services; the Clearinghouse
for Network Information Discovery and Retrieval (CNIDR); and the
InterNIC, which provides directory and database, registration of
domain names and information services.

NSFNET program solicitation (1993) - A document (NSF 93-52) that
describes new services and architecture for the NSFNET program.
it invites proposals for four separate projects: regional network
providers, network access point managers, routing arbiters, and
very high-speed backbone network services.

OC-3 - Fiber optic network standard. Transmission rate of 155
Megabits per second.

OC-12 - Fiber optic network running at 622 Megabits per second.

parallel computing (or parallel processing) - The simultaneous
application of more than one processing unit to a single
computational problem. It approaches a complex problem by
addressing its individual components simultaneously and
integrating the results. Parallel processing is the focus of much
research because it holds promise for immensely improving the
ability to solve complex algorithms and to quickly evaluate vast
amounts of data.

regional network providers - local networks. These networks
formerly provided the local links between universities and the
NSFNET backbone--the regional network providers now have national
access through Network Service Providers.

routing arbiters - promote Internet routing and stability,
establish network topology and policy databases, develop
procedures to resolve problems between network entities, develop
advance routing technologies, provide simplified routing
strategies, and promote distributed operation and management of
the Internet. The Routing Arbiters offer routing and other
services to facilitate the logical interconnection of networks.
Two five-year collaborative cooperative agreements have been made
to offer routing arbiter services: Merit Network, Inc. and the
University of Southern California's Information Sciences
Institute.

T1 circuits - transmission circuits introduced to the NSFNET
backbone in 1988 which transmit data at a rate of 1.544 megabits
per second (Mbps).

T3 circuits -  transmission circuits that superseded the T1
circuits in 1991, providing the NSFNET backbone with a 45-Mbps
capacity for data transmission.

very high speed backbone network services (vBNS) - An advanced
experimental network provided for through a cooperative agreement
with MCI (initially 155 Mbps and anticipated to grow to gigabit-
per-second transmission) connecting the NSF-funded supercomputer
centers, and allowing for research on high-speed networking. this
network is not an operating infrastructure; regular Internet
traffic will not travel over this virtual network. Research using
the vBNS may later be developed into technology for use by the
various commercial service providers.
------ end of nsfmci.txt -- ascii -- complete ------



More information about the Methods mailing list