[Internet 2 -- More reason we need xDSL NOW]
Getting Ready For Internet2
(08/18/97; 9:06 a.m. EDT)
By Larry Lange, Electronics Engineering Times
ANN ARBOR, Mich. -- The effort to build Internet2--an academic
and government research-only network separate from the increasingly
commercialized Internet--is kicking into high gear following a series
of technical meetings this summer.
The effort is rapidly picking up participants from universities that
want to log on to the proposed high-speed, multimedia-oriented
network. And communications OEMs are eyeing Internet2 as an ideal
test bed for their next-generation systems.
Having completed a critical technical workshop here early this month
and one in July in Denver, the Internet2 organization is pushing to
meet its self-imposed deadlines for bringing its network live by year's
end, with three university members to be connected as early as this
month.
The network promises benefits for commercial as well as academic
Internet users. Obviously, offloading institutional traffic from the
commercial Internet will speed transmissions for both Internet and
Internet2 users.
But the "I2" will also "speed the development of new network features
and functionality," said Charles Lee, senior manager of Internet2
interests for MCI Communications Inc. Those features and functions
"could then be incorporated into the commercial Internet," Lee said.
The work is coming at a time when the underlying base technology of
the Internet Protocol is being pulled in many directions. In this
environment, communications OEMs are eager for a safe haven to test
out future technologies.
"Internet2 will provide a large-scale environment to try out new ideas
and equipment running at speeds faster than any in operation with an
intellectual community of users," said Stephen Wolff, director of
business development for Cisco Systems Inc., a participant in the
project.
The Internet2 organization has grown from 34 members at its October
inception to more than 100 research universities today. Notable
government representatives hail from the U.S. Department of Energy
and the National Science Foundation (NSF, Arlington, Va.). Several
members are already engineering prototype network facilities, with
the project's initial operations expected to begin by year's end.
"We originally thought only 20 to 30 of the major research universities
had such a high priority for continuing to build their network
infrastructure that they would join the project," said Internet2 director
Mike Roberts of Educom, a consortium of universities that promotes
the use of information technology in education and research.
"Obviously, we were wrong."
So far, Internet2 university members have committed up to $50
million per year in new funding for the project, and the commercial
sector--with project representation from such companies as Cisco,
Lucent Technologies, Sun Microsystems, IBM, AT&T and
Microsoft--has pledged more than $5 million. By contributing
funding to university partners, the corporations can tap the project as a
testbed for unproven Internet technologies--a task that's difficult to
undertake on today's overcrowded Internet.
Internet2 organizers are working with representatives from IBM, MCI,
Cisco and other internetworking companies to discuss the needs of the
system architects developing the advanced Internet2 nodes called
Gigapops (points of presence)--the network aggregation points for the
second-generation Internet wide-area network.
Ted Hanss, director of applications development for the Internet2
project, told EE Times that such applications-enabling technologies as
security, multicast and quality of service were among the important
topics discussed at the summer meetings. The applications heading the
list for I2 are collaborative environments, digital libraries,
tele-immersion, tele-medicine and distance-independent instruction,
Hanss said.
Internet2 engineers have officially settled on an architecture of
connectivity through Gigapops, which can swiftly connect the
campuses, labs and, later, the urban-area and state/regional networks
envisioned by the Clinton administration's Next-Generation Internet
(NGI) initiative.
Accordingly, Internet2 engineers look to deliver 622-Mbit/second
transmissions to three supercomputer centers, running the Internet
Protocol (IP) over an asynchronous-transfer-mode (ATM) network.
And they want to boost bandwidth by nearly 14 percent for the
University of Illinois National Center for Supercomputer Applications
(NCSA), the San Diego Supercomputer Center and the Pittsburgh
Supercomputer Center, with the promise of 20 more universities to be
linked by this fall to Gigapops that will deliver data at speeds of at
least 155 Mbits/s. (Current commercial transmission speeds top out at
45 Mbits/s.)
Other Internet2 project members have laid the foundations for
connectivity to the network based on its anticipated speeds. A group of
universities in California recently announced the building of an
Internet2 network that will be designed to connect campuses at speeds
of over 600 Mbits/s.
"The electronic highway is faced with rush-hour traffic most of the
day. We need reliable service delivery," said M. Stuart Lynn, associate
vice president at the University of California and the principal
investigator for the project.
The participating schools include seven campuses of the University of
California, along with the California Institute of Technology,
California State University, Stanford University and the University of
Southern California. Notable goals for the project will be the
formation of a virtual university in which students can view
publications from distant libraries and take classes located at other
campuses.
Internet2 is systematically swallowing up the National Science
Foundation's Very High-Performance Backbone Network Service
(vBNS). More than 50 Internet2 institutions have received
competitively awarded vBNS grants under the NSF's High
Performance Connections program.
In fact, vBNS could be considered the heart of Internet2, or at least its
substantive launchpad. Begun in 1995, with an investment of $50
million under a five-year cooperative project with MCI, the service
links six NSF supercomputer centers and was initially implemented to
design and support "gigabit testbeds" for R&D of advanced
networking technologies. The centers are located at the Cornell Theory
Center, at NCSA and elsewhere.
Those technologies included ATM/Sonet, the interfacing of ATM to
the High Performance Parallel Interface and HiPPI switches, and
all-optical networking. Each testbed addressed an application that
required gigabit-speed networks.
The trunk-line infrastructure for true broadband services to academia
is being defined at a time when the core protocols for the Internet are
being upgraded from the original transport and network protocols
developed by the Advanced Research Projects Agency (now Darpa) in
the 1980s.
NSF program director Mark Luker said many of the problems
restricting the commercialization of the original Internet "should be
solved within the next few years using Internet2." Once traditional
not-for-profit users move to the new network, Luker said, "it will be
easier to implement pay schemes and give incentives to finance further
Internet growth."
Luker's comments underscore one of the motivations behind the
administration's NGI initiative: relieving the current Internet's
bandwidth bottlenecks.
"This is an ongoing proposition. It won't stop with Internet2. I see an
Internet3 and 4 in the near future," Luker said. By the end of 1998,
Internet2 looks to have nearly all participating universities employing
stable Internet2 connections.
Projects under development with Internet2, Luker said, include digital
multimedia libraries accessible in virtual reality, enhanced
collaborative workplace communities with live digital video feeds,
videoconferencing, collaborative computer-integrated manufacturing,
weather forecasting and military-troop-movement monitoring.
A briefing on Internet2 by the NASA Ames Research Center at
Moffett Field (Mountain View, Calif.) drew more than 60 companies.
"Although much of the research needed to make a new, superfast
Internet is too risky and long-term for the private sector, our success
will depend on partnerships with private industry and universities,"
Christine Falsetti, NGI project manager at Ames, said.
NASA recently designated Ames the lead institution for the agency's
$30 million portion of a three-year, $300 million federal project to
develop the NGI. "We want a network for researchers that is fast from
end to end. And we will work with private companies on routers,
switchers and computer workstations that will send computer
information much faster than today's machines can send it," Falsetti
said.
One NASA goal is to create "co-laboratories" by linking labs,
computers, databases and scientists from around the world via I2.
Ames is organizing a September workshop for companies interested in
such emerging applications.
The NGI initiative was unveiled in October with three basic goals: to
connect universities and national labs with high-speed networks that
would be 100 to 1,000 times faster than the commercial Internet; to
promote experimentation with advanced multimedia technologies,
such as real-time videoconferencing; and to demonstrate new
applications that support scientific research, national security, distance
education, environmental monitoring and health care.
On the campaign trail, the Clinton administration promised $100
million in funding for its initiative. The funds are set for release in
January.
Internet2 spokesmen are quick to say that the network is not designed
to replace the existing public Internet or to sidestep the NGI initiative.
"The goals of Internet2 and of the NGI are entirely compatible and
complementary," said Douglas Van Houweling, vice chairman of
Internet2. "There is a compelling unanimity of purpose and direction."
Yet Internet2 is not without its critics. There have been suggestions,
for instance, that universities are leveraging the project to maintain a
separate network presence.
But Internet2 Steering Committee member Raman Khanna disputed
that assertion. "At Stanford, where I work, only 15 percent of Internet
traffic goes toward other universities," Khanna said. "The other 85
percent connects the school to the rest of the commercial network.
Even if academia were to isolate its future, high-speed network
system, it would speed up only 15 percent of its Internet traffic."
Khanna also noted that the high-speed network is merely for the
development of advanced applications and that any advances made will
be available to all Internet users in three to five years.
Another point of controversy is the government's role in funding
specific university projects though Internet2. Several high-level
lawmakers complained at recent Senate hearings that the NGI program
favors urban areas and large universities.
But Neal Lane, director of the National Science Foundation, said that
Internet2 "is not an established infrastructure like a highway. It is a
work in progress . . . and it is experimental.
"That is the reason we need the nation's researchers to help us move
forward."
It remains to be seen whether the Internet2 body can pull off its
monumental project and then seamlessly bring the network into line
with the commercial Internet and provide bandwidth and multimedia
solutions for the NGI, all by the slated deadline of 2000. If it can, the
project will look to realize even more impressive goals.
Internet2 plans to share discoveries with others in education
worldwide. "This is the approach that characterized the first Internet,"
said NSF's Luker, "and it can work again with Internet2." |
