![]()
|
|
||
| Volume 1, Number 21 | April 15, 1998 | home / experts / internet |
|
|
||
Internet2 and Abilene: Decoding the Hype |
A National Network Faster Than Most Campus Networks?
Wow. So will people actually send the entire contents of the Library of Congress over this new network? No, silly. The contents of the Library of Congress haven’t been digitized. While someday in the far future they may be, for now, people just like to use that as an example of how fast the network will be. (And no one knows how to check the veracity of the claim.) In the real world, Abilene will be used to access materials in existing digital libraries, as well as for many other kinds of applications. Such as? Well, there are a number of digital library projects under way, putting large quantities of text, still images, sounds, and even video online. A Web-based course reading list at the University of Texas could rely on materials from the Berkeley digital library, for instance. We can also expect this network to enable much more in the way of virtual lecturing by national experts in various disciplines – both real-time lecturing, and playback from digital archives from the greatest professors’ best lectures. Finally, there are all sorts of research applications: modeling airflow over aircraft wings, modeling the weather, collaborating on medical research, etc. Wait a minute, didn’t I hear that exact same list of applications in the mid-80s when the NSFnet was created? Touché! But think about how much RAM and disk space you had in your PC ten years ago. Now all the fancy applications in use at universities are much more sophisticated, so we need more bandwidth to move the data around. Plus, these days we think more in terms of real-time applications. The classic example cited by Internet2 officials is telemedicine: for instance, a real-time video feed of a surgery in progress so that a specialist in another part of the country can consult. We couldn’t do that sort of thing at the speeds of the old NSFnet. Another important distinction between Abilene and the NSFnet lies in where the computing will take place. In 1988, there were a handful of supercomputer centers, and researchers across the nation were expected to use the NSFnet to submit data and retrieve results from that small group of special sites. Today, the market for supercomputers is virtually dead, as small clusters of fast boxes – workstations such as DEC Alpha boxes, PowerPC servers, even Pentium II PCs in clusters – routinely perform what we used to call supercomputing. Computing is much more democratic and much more distributed. How does this network capacity compare to today’s campus networks? I’m glad you asked that. In theory, this national backbone will be at least as fast as the networks in use today on university campuses. Building networks commonly run at 10 megabits per second or 100 megabits / second. Most large universities have fiber links running at several hundred megabits / second, and some university laboratories are testing speeds faster than that. So this means, in theory, that we could think of the Abilene network as one gigantic campus network, with sufficient bandwidth to allow students and faculty to treat remote resources as if they resided on a single campus. Think of having the performance of a LAN across all the major universities in the country. This could change the very nature of universities in the U.S. More... |
Comments are welcome
Produced by Rich Wiggins and
All Rights Reserved. Legal Notices.
Created: April 15, 1998
Revised: April 15, 1998
URL: http://webreference.com/outlook/column21/page2.html