A bump in the road might be ahead, but look at this, then settle back and enjoy the long ride:
Dow Jones Newswires -- December 2, 1997
Researchers Living In Future With Next
Version Of Internet
By JOELLE TESSLER
Dow Jones Newswires
NEW YORK -- With the Internet of the 21st century, doctors will be able
to consult with patients and other physicians who are hundreds of miles
away.
Among other things, "telemedicine" will use advanced computer networks to
transmit X-rays and other images, to observe eye vibrations during
neurological exams and to remotely control medical instruments.
Meanwhile, scientists at four atmospheric research institutions in the
Western U.S. are building on the next version of the Internet to develop a
"distributed modeling laboratory" to conduct regional meteorological studies.
Advanced networks will join high-performance computers at the four
institutions to create a laboratory to operate weather simulations, perform
real-time forecasting and other experiments.
Even as the current Internet continues to grow as more and more homes and
businesses hook up to the Web, government, university and corporate
researchers are laying the foundation for the next version of cyberspace.
Several interconnected initiatives are building advanced computer networks
that will be faster, more reliable and smarter than today's Internet. And with
these capabilities, these new networks will be able to handle applications
that aren't feasible on the current Internet.
"We want to allow some portion of the research community to live in the
future because all of us will eventually have access to networks at these
speeds," said Tom Kalil, senior director at the National Economic Council in
the White House.
The Clinton administration has spearheaded the advanced networking
projects with its Next Generation Internet Initiative, which the president
announced in October 1996.
NGI, as the top-down project has come to be known, has set three main
goals, explained Sally Howe, acting director of the National Coordination
Office for Computing, Information and Communications, which supports
NGI.
The first goal is to research and develop the technology needed for the next
generation of the Internet.
A key objective is to develop "differentiated service," which will provide
different levels of service for different pieces of data. Differentiated service
will, for instance, assign certain data priority to be sent at fast speeds,
without delays and at steady transmission rates.
"One of the biggest problems with the existing Internet is that people want
better service than they can get," said Stephen Wolff, executive director of
the advanced Internet initiatives division at Cisco Systems Inc. (CSCO).
"No matter how much money you have (to invest in technology), all traffic
goes into the same congested swamp."
Today's "best effort service" forces everyone to share the same bandwidth,
which can cause significant transmission delays and jerky images due to
fluctuating delivery rates. While this may may not affect e-mail, it makes
applications like video transmission and multicasting - sending video and
other data to many people in different places - difficult.
Differentiated service will, in essence, create "information passing lanes" for
certain types of data, said Charles Lee, program manager and executive
manager in MCI Communications Corp.'s (MCIC) federal branch.
Differentiated service will also provide different levels security and privacy,
among other features.
The intention, Kalil stressed, is to create "not only... high-speed networks
but also smarter networks."
NGI's second goal is to set up actual high-performance networks to test and
demonstrate the technology being developed. The objective is to connect at
least 100 sites to networks that are at least 100 times faster end-to-end than
today's Internet, and to connect about 10 sites to networks that are 1000
times faster end-to-end, Howe said.
To do this, NGI is building on and upgrading high- performance testbed
networks at several federal agencies.
NGI is working to set up the 100 sites with the National Science
Foundation's very high performance Backbone Network Service, or vBNS;
NASA's Research and Education Network, the NREN; the Department of
Energy's Energy Sciences Network, or ESnet; and the Defense
Department's Defense Research and Engineering Network, or DREN.
Although the testbed networks existed before the president's initiative, "NGI
brought the agencies working separately and asked them to work together...
interconnecting separate networks," said Mark Luker, program director for
high-performance networking at NSF.
To build the 10 sites that are 1000 times faster, NGI is working with several
experimental networks set up by the Defense Department's research and
development arm, the Defense Advanced Research Projects Agency, or
DARPA.
The final goal of NGI is to develop revolutionary applications that will make
efficient use of this new technology and these high-performance networks.
To permit applications that require more bandwidth than the current Internet
can offer, the testbed networks may allocate usage so that applications
share bandwidth with only a limited number of other users. This strategy
makes more sense than just adding bandwidth since additional capacity can
quickly become clogged.
Since these networks have fewer users than today's Internet, they will also
provide a clearer path for data. This will increase transmission speeds and
reduce latency times - the time it takes to influence a remote computer and
get a response - to allow applications that require faster networking than the
commercial Internet can provide.
"The analogy is an eight-lane highway - more traffic slows you down," said
Larry Smarr, director of the National Center for Supercomputing
Applications at the University of Illinois, which is one of the backbone sites
for the vBNS.
The list of potential applications is quite long - particularly since, as Howe
pointed out, there are "things we haven't even thought about."
To allow applications like telemedicine and collaborative research, to name
just a few examples, these networks will enable users located miles apart to
share virtual space and permit researchers to remotely control instruments
like telescopes and medical devices.
Beyond the federal agencies that run the networks and university-based
researchers working with them, the National Library of Medicine and the
National Institute of Standards and Technology, or NIST, will develop
applications.
The National Library of Medicine, part of the National Institutes of Health,
or NIH, will promote medical applications like telemedicine. NIH envisions
a world in which doctors can use telemedicine to treat people who cannot
get to a doctor easily - like the elderly - and patients who do not have local
access to the specialists they need.
NIST, which is part of the Commerce Department, will study applications
related to manufacturing. NIST will also study the standards that are
necessary to make the separate networks compatible.
Also, NSF is awarding grants to about 100 universities with deserving
applications to connect to the vBNS.
The vBNS, built and run by MCI, is probably the best-known and the
largest of the NGI initiatives.
The network, which became operational in April 1995, transmits at 622
megabits per second on the backbone. By comparison, the average
multibillion dollar corporation probably has a network that transmits at no
better than 45 megabits per second, Smarr said.
NSF first set up the vBNS to operate among the five NSF supercomputing
centers - including the National Center for Supercomputing Applications -
established in the mid-1980s.
About a year ago, however, the agency decided to focus more on
developing applications over the network. NSF therefore upgraded the
backbone, added more access nodes and began working with universities to
connect to the network. More than 20 universities are now up and running
on the vBNS and that number will be about 40 before the end of the year.
Meanwhile, most of the universities that are linked to the vBNS or hope to
link to the network have formed a separate but overlapping organization
called Internet2 to study educational applications over high-speed networks.
The group "will focus more on distance learning and things that universities
care about," the NSF's Luker said.
At last count, Internet2 had 116 university members.
Although the organization plans to work with networks other than just the
vBNS, it is using the vBNS as its initial backbone and is currently working
to establish a more formal relationship with NSF, said Douglas van
Houweling, president and chief executive of the University Corporation for
Advanced Internet Development, which oversees Internet2.
Clearly, the cost of researching and building these networks and advanced
applications isn't cheap. While Clinton had initially pledged $100 million a
year for three years for these projects, the federal funding has faced intense
congressional scrutiny.
Congress has appropriated $80 million of the $100 million that the Clinton
administration had requested for NGI for fiscal 1998, plus another another
$5 million for NIH.
The administration will distribute these funds to NSF, NIH, NASA, the
Defense Department and NIST. Although the Energy Department did not
receive any new money this year, the agency will still work on related
research.
The federal agencies are also using other federal funds to finance these
projects.
NSF in turn is awarding a total of about $10 million a year - before another
$10 million out of the $23 million in NGI funding that it expects to receive in
fiscal 1998 - to universities to link to the vBNS.
But these grants, which come out to $350,000 over two years each, are
really only seed money, Luker explained. The universities themselves must at
least match that money.
And Internet2 schools have committed to each spend at least $500,000 a
year - although the actual cost could come closer to $1 to $2 million a year
- to build the campus network capacity and the connections to hook up to
the vBNS and other networks.
Industry partners are also investing large sums. NSF pays MCI about $10
million a year to build and run the vBNS. But the network "costs maybe 10
times that," with MCI's investment "in the hundreds of millions," MCI's Lee
said.
Likewise, Cisco offers $100,000 in equipment, including high-end routers
and switches, to Internet2 institutions with NSF grants, and has offered
discounts on equipment to all Internet2 institutions. The company is also
submitting proposals in conjunction with universities to federal agencies
participating in NGI to take part in their projects.
"The $85 million will leverage a lot of additional money from industry and
universities," said Kalil of the National Economic Council. "The government
is acting as a catalyst."
Both Lee and Cisco's Wolff stressed that their companies see their
involvement as an investment in the future of their own businesses since they
hope to ultimately bring this new technology to the commercial market.
Many companies are therefore providing employees as well as equipment
and money to form an "intellectual partnership," as Wolff put it.
"We think it's a good investment," Lee said. "Knowledge about how this is
done is pretty valuable."
Indeed, Luker noted, participating in these types of projects will help these
companies remain competitive. "The industry realizes that the technology is
not done yet," he said. "There is a strong push to get to the next
technology... and it really matters who gets there first. This is the best way to
be on the leading edge."
"We have observed that the research and education communities often
provide a leading indicator of the commercial market," Wolff added. "We're
learning from them too." And that is the basic theme underlying all of the
initiatives to develop the next version of the Internet.
Although government and university researchers are developing these
advanced networks and will be the first to use them, the technology will be
the foundation of the next version of the commercial Internet.
"The larger community will demand this technology," Wolff said. "We will
see some fairly wonderful things come ... to market within two to three
years."
In fact, today's Internet began in much the same way as networks run by
DARPA and NSF entered the public domain, as companies commercialized
the technology used to build these networks, and as the students, professors
and researchers working on them started their own companies.
"The goal is to invest in research and development and network testbed that
will create the foundation for the 21st Century in the same way that
(DARPA's) ARPANET and (NFS's) NSFNET led to today's Internet,"
Kalil said.
So while it is hard to predict for certain what the next version of the Internet
will look like since the technology is still evolving, one thing seems clear.
"The Internet will be different," Howe said. |
