Eugene, Oregon
May 13-15, 1998
Mark Luker, Vice-President
Educom
Thank you. I'm very pleased to be here with you today to focus on the National Science Foundation's (NSF) role in the Next Generation Internet Initiative (NGI). And since I now work for Educom, one of ARL's major partners, and we are merging with another one of your major partners, CAUSE, I would like to devote a little time to that merger and the merger within a merger that may impact how the higher education community addresses networking, as well as some time on how we can work more closely with ARL.
You've all heard that Internet years are something like seven to one compared to normal years. The old NSFnet backbone was the Internet for the world only about five years ago. It was the stage after the experimental Net, and was the first wide-area national scale network of this type. It proved that a great deal of things were possible, whetting a lot of appetites. The world has changed a lot in five years. (See slides 1, 2, and 3.)
When I joined NSF about 2-1/2 years ago, the Foundation had started a new program called "A New Architecture" and awarded some grants to build network access points or meeting points where new, fledgling, commercial networks could come together and exchange information. They built something called the routing arbiter, a way to find out how to send information to the right place. And all this was to set up an environment for free market competition and the delivery of packets according to Internet protocol.
What happened is that all went faster than NSF predicted. So when I got in on what was a fairly new program, my original appointment was to wind that down, declare victory on the commercial Internet, and move on. We finished some 18 months early in that case.
What NSF then did was turn its focus back to its original customers--the research and education community--and see what might be done in education or in networking to meet the special needs of that community.
This is when NSF started the vBNS National Network Program. This was a particular physical network run by MCI on behalf of NSF. It was promoted on the basis of a national competition and originally connected five supercomputer sites to each other. Again, in very recent history, that's advanced networking for the world. (See slides 5, 6, 7, and 8.)
The goals of the program were good in principle, and they have lasted right through the present day into the Internet2 project and the new program with MCI. One is to support national and global scale research and education. This is an infrastructure project: Support the research and education efforts of others by providing a platform. This is one kind of activity in NSF similar to building a large, national accelerator so the physicists can do their work better. The other model, of course, is to give the money straight to physicists to do their work. NSF is always treading a fairly narrow line between spending more money on infrastructure for community projects and spending more money directly on the projects themselves.
It is recognized now that we need to look at a new kind of Internet technology, a multi-lane highway with traffic controls and possibly tolls. That was a technology never imagined in the first Internet and one that requires fundamental core changes.
The first typology of vBNS, as I said, went to just a few sites, but rapidly grew under a program that managed to last two years and connected about 100 universities to the network. We went from five to about 100 nodes.
As you may have heard, we were having interagency warfare over who controls advanced networking. That was going on in the background. And the various marketing wings of Sprint, MCI, and others had a contest to see who could produce the map with the most lines and arrows.
It's a sign of adolescence in the technology, actually, when you can draw a map of your network and talk about who is connected and who isn't and how big the pipes are from here to there. That means you're dealing with a fairly small-scale project. And this is small when I talk about 100 nodes. You don't typically see maps that cover the entire United States of the global telephone system, for example. We know in this country, at least, that it goes basically everywhere, and the map is superfluous. I don't think we'll see very many more maps of advanced networking because of expanding developments such as the Abilene project. We are getting more and more competition and new fibers in the ground.
One of the things that came out of this project with vBNS was an award to 93 universities. There NSF followed its original principles: They did not simply connect the university to a network, the universities had to form a partnership inside to figure out what applications they would use. So going into this, we have about a 100 times ten or so well thought-out applications, not all of which can be used yet because the connections with the network aren't there. But the university faculty, research staff, and others had to demonstrate in advance they had a good use for this thing before they could ask for it. I believe the actual number of awards to date is 92, and something like half of them are connected.
The other partnership universities had to establish was with their local telephone, fiber, or power company. This has turned out to be very difficult. You cannot go down to your local telephone company and order one of these OC-48 connections to the nearest big node on the network. They don't sell those things. So part of this, of course, is to force private industry to move a little quicker. Many of the big companies, for instance, had this kind of thing sitting in their research labs, but had not moved it into their marketing organization yet. They had sort of gotten caught up in figuring out how much it would cost--this is inside the company--and what kind of new contracts their lawyers should work on to handle this new kind of service.
These are not minor points because the kind of network we are describing here can subsume the entire telephone network. It can handle all of the voice needs easily. It can subsume the video network easily. So the various legal and contractual structures you have built around those worlds all come together in the same room if you talk about convergence. Those issues are real. NSF played a good role by trying to make that happen sooner. I'm sure many of you can tell war stories about how it is to make the right kind of legal and contractual connections to get high-speed networking to the campus.
At the same time, NSF picked up a number of partners both in industry and especially in the other federal agencies. This is when the Next Generation Internet project started to form on an informal basis. NGI is a project involving quite a few different government agencies and especially the ones that have big research networks in the country: e.g., The Department of Energy, NASA, and DARPA. Each of these have been operating in the vertical stovepipe mode: Building their own networks, assuring that their maps had more spots than the other maps, and basically competing with each other over the same kind of service. This project has the very good effect of starting to interconnect those networks physically and getting the engineers together in the same room so they could start to cooperate even though the marketing people and contract officers couldn't.
STAR TAP (Science, Technology and Research Transit Access Point) is another award from NSF which simply provided a platform like the original network access point, but one for the global network. It ended up at Chicago for historical, but important, reasons. This again shows, however, the political nature of all of this: Why have only one STAR TAP, one connection point, in the United States?
Well, actually, nations that bring in networks can connect in a variety of places. The reasons to designate one is so that we could force, if necessary, networks from the Pacific Rim countries to come to the same place as the networks from Europe. Why? If they came to the West Coast and the East Coast, someone would have to pay for the connection between them. Our Federal Government won't pay for a connection for Germany to talk to Singapore unless it involves some of our research or education activities directly. So there are commercial issues such as transport and who pays for it. The STAR TAP got around that, but it's not over yet. And the various core networks are coming quite regularly to the United States to ask us to pay a much larger share of the so-called transoceanic rates.
That these major projects became formalized is a development that can't be overemphasized. First of all, the Next Generation Internet became a formal federal project. It was announced by Vice-President Gore and President Clinton as part of a quick campaign stop, and then became a very rushed job for the federal agencies to define what it meant, because it was only a few months between the campaign stop and when Congress was considering funding. I was lucky to serve on the committee that wrote the plan for the NGI during that time. We had to quit talking about it and actually do it, try to write a plan that incorporates all the agencies. The NSF vBNS program went into that plan as a major infrastructure piece that brought the universities in directly. Today they form a key part of the NGI project.
The other big development--and I think the bigger one by far in the long run--is the formalization of the Internet2 project and the incorporation of UCAID. This organization has really energized the universities, the vendors, the public, and the government around this topic; it has provided a great focus and, with the Abilene Project and others, it is bringing in entirely new networks.
These networks are different. The Abilene Project is built on newer technology than the vBNS. We are hearing of basic changes that will affect the cost performance of networks by orders of magnitude. Others are coming down the road. So I think we are just reaching the stage that we have always dreamt about, having true competition in the development of these advanced networks. The marketplace is starting to work.
George Strawn (NSF) and Dave Nelson (DoE) of the Next Generation Internet Project always present an organization chart of how the federal government deals with advanced networking and advanced computing. But since they reorganize and change their names more than once a year, it's a difficult thing to track. I was able to take their slides showing the organization and make the chart a straight line (see slide 11). You can trace the line from the White House--I believe it's eight layers deep--to the NGI project, which has a single point of leadership. Before, it was almost a club, a large number of boxes loosely coupled to each other.
The NGI legislation under proposal right now adds up to $110 million dollars a year. This would put $25 million dollars more per year into the National Science Foundation. A lot of that would come to advanced networking infrastructure, the kind of thing that supported the vBNS in the past. The real question before us now is what it should support in the future.
What are the major issues at NSF with the NGI? The first one is: Should the federal government continue to support infrastructure programs for advanced networking? Most of my communities have a very kneejerk answer to that, of course, as do many in the university community. But I will say it is a real argument every year. Much of industry says that industry should do this and not government, so they don't all accept the fact that there's a role for the federal government to push the leading edge.
Even when you get that far, there's an open argument every single year about whether the money should be put in infrastructure, in a big national platform, or in networking that supports the research and applications for everyone. Or should you put it into computer science research, into further-out concepts, blackboard and chalk research? And what might be the network of the 22nd Century? Those are open discussions, and they lead to a balancing act every year.
If the federal government is going to support infrastructure, how best to do it? It's more complicated this time. The vBNS program is now about three and a half years old, and five years back there was no large-scale networking project, and there was no commercial industry that could provide much of any part of it, and there was quite a bit of unanimity on where to go. Right now there are new technologies, new networks, and new companies. The world is dramatically more complicated. How do you operate an advanced networking program in that kind of world?
There's real competition now both in the business of networking and in ideas about networking. NSF, the government, and Congress in general are firmly on the side of free market to figure out what to do in the case of many options like this. But how do you operate such a thing for an advanced network when it also has to provide quality of service from one end to the other? The technological requirements require a lot of coordination, possibly through standards only, but maybe needing more, yet the free market demands a lot of competition.
There are many more stakeholders. Everyone cares about this now and has strong, if not violent, opinions when it comes time to talk about how to spend government dollars. This slide has been used many times to illustrate why the federal government should be involved, and it really reflects the history of networking at several different levels, starting with research projects, moving into infrastructure projects supported by the government--such as NSF's--in places where the commercial model is not yet ready, and then moving on into private industry.
Now let me talk about organizations because I think the merger of Educom and CAUSE has something to do with all of this. Of course, Educom, CAUSE, and ARL have been the three major organizations working on quite a few joint projects, including CNI, for example. Educom and CAUSE are, of the three, the most alike. They came from basically information technology routes, not information routes. Though we started at somewhat different places, we have been networked, as they say, and are converging.
The digital convergence that brings all applications together in the same technology eventually does that to organizations as well. The whole role of the CIO on campuses in the last few years has been to accomplish that: Reorganize to bring together what were separate organizations in order to provide a single, common service. So Educom and CAUSE are practicing what we preach by merging. Further, because our programs were complimentary, the merger is fairly pain-free and will immediately bring about a stronger program.
We will represent about 1,600 campuses--that means all shapes and all sizes. One common feature, of course, is they all need better networking. In fact, they all need advanced networking as soon as possible. The major players in the Internet2, for example, are out there leading the charge, but the Internet as the great equalizer is required at all the smallest, most remote campuses as well, where it creates an even bigger effect there.
We will, of course, be able to speak with one voice instead of two and not have to explain to our audience the difference between Educom and CAUSE. I might say I've spent quite a bit of the last two years trying to explain the difference between Internet2 and NGI. I'm glad that we are starting to get past that point. We really are looking at different sides of the same beast, and it's not productive to spend a lot of public relations time in Congress or elsewhere talking about our differences when our goals are exactly the same.
I want to mention three projects, then, that came to the new organization, EDUCAUSE, out of Educom. The first is NLII, the National Learning Infrastructure Initiative. This is a very vibrant organization that has been alive for several years now. It is focused on educational applications for networking, aiming to provide affordable, quality access to higher education, via the Internet. The quality here is defined not as much as networking quality, but educational quality.
Affordability is the key issue. In the past, we always worried about whether we could do something in the network and the computer without thinking about how much it cost. There had been a lot of good demonstrations of nice things, but when it came time to use them at our universities for all of our students or even all of one class, we weren't able to afford it. The NLII focuses on affordability as well as use.
Now, once you do this, whether it's for your students on campus or at a distance, the specter immediately rises of how to provide all the other services to go with education. It's not just instruction, it's not just tutoring--it's libraries. That one comes to the top right away. So if you're going to use this in some kind of asynchronous, electronic network way, you have to provide similarly high-powered, quality access routes to library materials and all the other student services, right down to registration and everything else.
The Instructional Management System is a big project now, with a multi-million dollar budget with many industrial partners. It is a project trying to create standards that can be used nationally or worldwide to allow the industrial model or market model to develop instructional software. It incorporates all aspects that we have talked about for library access in the past.
The second project that comes to EDUCAUSE from Educom is the Coalition for Networked Information. You know a lot about CNI, so I'll just say that it is a very important enterprise from the viewpoint of EDUCAUSE. A word about networking issues. One of the CNI projects is working on middleware, network services such as security and authentication, all that it takes to do commerce in intellectual property, whether it's for money or for free. It's moving the sites up one level from basic connections, which is what NSF is all about, into network services.
Finally, a word about NTTF, the Networking and Telecommunications Task Force, formerly of Educom. This was an organization of the CIOs who built the original campus network working with NSF. Doug Van Houweling and his comrades were key players in the NTTF. It was the incubator for the Internet2 project and helped form the UCAID Corporation. I'm very proud of that from an organizational viewpoint. That's a large win for the NTTF. But then the natural question, of course, is what's next, because the advanced networking issues naturally moved into UCAID. This is their home now. What will NTTF do next?
One thing we have done is a merger within a merger. We have just merged with FARNET. FARNET is an organization comprised of the directors of state networks. These state networks go back historically to those 13 nodes on the old NSTnet Backbone. They built the first regional networks for the Internet. Today they are networks that are usually confined to state boundaries. They usually connect the major universities in the state, but also the private colleges, smaller universities, libraries, public libraries, and businesses. In short, they are usually government, non-profit research and education networks, but at a community or regional level. We are bringing these directors into EDUCAUSE. I might say, many of them are very advanced. They are working with things called gigaPoPs, the main connection points with the Internet2 project. But in terms of geographic scale, they are the regional connections. For matters that are important to us, such as getting libraries connected to play direct roles in these applications, they will be able to provide us with some connections, not just network connections but also the human networking connections, the consortium.
FARNET has been a consistent partner of NTTF. NTTF helped NSF figure out what to do, usually by holding workshops or other activities that would bring the community together, study, and make recommendations. Now I think there's great synergy here. Part of it comes from the CIO vantage point. The CIOs need to get their campuses connected globally and nationwide, the faster the better. This is part of the Internet2 emphasis. But they also need to turn around and use networking to support all consortia on campus, in the city, and in the region; to reach the students in their businesses and their homes, as well as on campus; and to work with you and your partners back home. This group now can represent both halves of that.
EDUCAUSE is going to be stronger in national policy, because we won't have to explain why we are different anymore. We both have been working directly on national networking policies, on the Alliance and other policy organizations, and now we are able to unify that work. The Alliance is very interesting. It's more of a coffee break discussion, but there we try to bring universities into consensus or into a common voice on some of the major policy issues. You're well aware of some of them. They're issues involving copyright, fair use, licensing, etc., that are really right at the core of what we do.
Some other issues are just interesting, but irritating. The NCAA recently has been supporting a bill that would force the university CIOs as well as the network service provider in general to modify their Internet in some undefined way to prevent student athletes from gambling at offshore websites. This group includes a representative of the NCAA who heads some serious discussion. But that kind of thing, when you think about it, gets right to the core of some of the issues for intellectual properties. It's sort of an unspecified prior censorship to prevent an action that might be a crime that may occur in the future. Just think about that in the broader censorship context.
Finally, we all work together on national seminars like NET98. In EDUCAUSE and in this merger we will be working with the Next Generation Internet people. That means the federal government side again. We have a grant to start a national program for outreach on advanced networking, a project that complements Internet2. Outreach means going beyond the universities that are directly participating and figuring out how much of I2 can be brought to the rest of higher education now, how to get the rest there as soon as possible, and which of its particular features are most important. This will really take advantage of the machinery we inherit from CAUSE, which has excellent development organizations. We'll bring all that to bear. FARNET has an excellent state network grant from NSF.
And finally, we have been asked to run another workshop on the next networking program at NSF, asking what NSF should do next. Again, thinking about Internet years, there's just a fraction of a year left on vBNS contract. Should they renew it and do it the same way? Should they recompute it? Should they get out of the business? Should they support market-based advanced networking in some other way, like giving the money directly to the universities and letting them figure it out? There are fundamental questions on how to define and configure grant programs when things are moving so rapidly. I'm excited by this prospect.
We will be working directly with CNI, ARL, and many others to advance the development of commercial networking. This again is a role where we represent a large number of colleges and universities. On the average, we will not get network service awarded to us free in the higher education community; we will have to buy it. Our goal, then, is for early and effective commercialization of the new technology.
I thank you and welcome your questions.