TechTalks Transcript
An Update on Internet2
 Judith Boettcher [JB] |
 Howard Strauss [HS] |
 Doug van Houweling [DVH] |
April 8, 1999
Audio
• Streaming
MP3
• Download
MP3 (Download
Tips)
JB: Welcome to the CREN TechTalk series for spring of 1999 and to this session on "An Update on Internet2 and What It Means to You." You are here because it's time to discuss the core technologies in your future.
This is Judith Boettcher, your CREN host for today, and I'm pleased to welcome our technology anchor for TechTalk, Howard Strauss of Princeton. Howard is a well-known Web and all-around Information Technology expert.
Welcome, Howard.
HS: Thank you, Judith.
I'm Howard Strauss, the technology anchor for the TechTalk series of technology Webcasts. The job of the technology anchor is to engage our guest expert in a lively technical dialogue that will answer the questions you'd like answered, and to ask those very important follow-up questions. You can ask our guest expert, Doug van Houweling, your own questions by sending e-mail to expert@cren.net anytime during this Webcast. If we don't get to your question during the Webcast, we'll provide an answer in the Webcast archives.
If you use the Web or e-mail, chances are you use the Internet. Is there still anyone who does not use the Internet today? Not anyone any of us know! We've become so unaware of the Internet, in fact, that we think of it as the applications that run on it. We don't say, "I'll sign onto the Internet tonight," but rather, "I'll check the Web" or "I'll check my e-mail."
There are many other applications on the Internet, but only people deep into networking and computing are aware of them. It is no surprise, then, that most of us think of Internet2 as just a faster way of doing e-mail on the Web. In fact, on our last Internet2 Webcast, we spent some time on the speedy VBNS (or Very-high-speed Backbone Network) and the speedier Abilene network that was on its way at the time. Now that Abilene is up and running, we will certainly bring you up to date on it.
But as much as we'd all like the Web to be faster, if Internet2 just delivered faster Web and e-mail access, it would hardly be something to write home about. The real purpose of Internet2 was never speedier communications. Its purpose was to enable new network applications, and it was those applications that required a much faster network.
On February 25, 1999, for example, the New York Times reported laparoscopic surgery being done using Internet2. Though you might not want to participate in or even watch such an event, the ability to control complex instruments remotely in real time is one of the many new applications that will be possible with Internet2.
Another is distributed middleware. Middleware? Didn't we cover that in our last Webcast? Well, I told you then that middleware was many things and could pop up in many strange places. On Internet2, there are many strange and unexpected beasts! Doug will help us sort out the critters in the Internet2 menagerie -- from the aardvarks to the zebras -- in today's Webcast of TechTalk.
Judith?
JB: So that's where that zebra comes in, Howard! Thanks very much.
And let me tell our guests -- our audience out there -- about our guest expert for today's TechTalk. Our guest expert is Doug van Houweling, the president and CEO of UCAID, which stands for the University Corporation for Advanced Internet Development. As most of us know (but need to be reminded now and again), UCAID is a consortium of US research universities in collaboration with private and public sector partners engaged in the Internet2 project. And the purpose of the Internet2 project is twofold: to advance networking technology and applications, and for the research and education community.
Doug is currently on leave from the University of Michigan. Prior to undertaking his responsibilities at UCAID, Dr. van Houweling was Dean for Academic Outreach and Vice-Provost for Information and Technology at the University of Michigan. As Dean for Academic Outreach, he was responsible for providing access to the University's learning environment, research activities and service programs, unconstrained by space and time. As Vice Provost for Information and Technology, he was responsible for the University's strategic direction in the Information Technology arena. Doug was also instrumental in the development of Merit, which is a statewide network in Michigan (nonprofit) that provides Internet networking for the universities in Michigan (and Doug may mention something more about that later as well).
Welcome, Doug, and thanks so much for being here. I imagine you have a lot on your plate getting ready for the Spring '99 meeting at the end of the month, and also with the recent launch of Abilene.
Doug, are you there?
HS: Doug, at least I can't hear you.
JB: I'm sorry.
HS: I can't hear him.
JB: No, I can't. You're very, very faint, Doug. Would you call back in again? Thank you.
Okay, Howard, we'll just tell our folks that are listening in that we had a very strange experience as we were doing the preparation for our talk today, and seemed to be getting an in-and-out experience here with Doug not being able to hear us all the time.
HS: Yeah, it's kind of interesting that, although we are talking about kind of an advanced technology -- namely, Internet2 -- that sometimes the technologies that we depend upon, like the telephone system, don't behave the way we'd like them to behave. But as soon as Doug dials back in, I think we're going to start.
Is that you, Doug?
DVH: It is.
HS: Okay. Well, we're going to start by perhaps asking you to bring us up to date on the current status of Abilene.
DVH: I'd be happy to. As I'm sure a lot of our listeners know, Abilene went into production in the end of January. And on February 24, here in Washington in Union Station, we had a launch event to mark the successful beginning of Abilene production. The Abilene network now has more than two dozen universities connected -- and more connecting all the time. Furthermore, the Abilene network is connected to other networks such as the Department of Energy's Energy Science Network and the NSF/MCI Worldcom-sponsored Very-high-speed Backbone Network Service.
HS: Doug, you said that there are two dozen universities connected right now and it's growing. How many do you ultimately expect to be on it?
DVH: It's very hard to know. We expect and hope that there will always be multiple high-speed backbone networks for Internet2, and we expect with around 150 members (which is how many universities we have joined right now) it could easily be that we would wind up with 75 or 80 connected institutions. It could be more than that, slightly less. Frankly, the key here is that the universities are able to acquire the high-speed networking services they need.
HS: And aren't there people on the Abilene in addition to universities? I understood there were some corporations and folks.
DVH: We will be providing connectivity to our corporate members' research laboratories and also to government laboratory facilities when and where they need it.
HS: Has Abilene completely replaced VBNS, or are there still folks on VBNS? What's the status of VBNS?
DVH: Well, indeed the VBNS is very much alive and well today. There are more universities, I'm sure, connected and doing their advanced Internet work over the VBNS than there are over Abilene, and it's my understanding that MCI Worldcom intends to continue providing the VBNS into the future. So we very much expect that there will always be at least two and probably more than two Internet2 backbone networks. We think that that helps us understand how to effectively bring multiple high-performance networks together in an operational environment.
HS: Okay, but since I pointed out in my little opening piece that the real importance of Internet2 was going to be applications, I assume that both the VBNS and Abilene will support the same applications -- or won't they?
DVH: That's our objective. In fact, we're working very hard with the folks at MCI Worldcom and with other people in the networks across the nation -- the federal networks -- to make sure that the facilities we build work transparently across the multiple network backbones they need to traverse.
JB: Will the same traffic go over both of the nets in terms of one using one network for part of the journey and then another part of the network for the other part?
DVH: That could easily happen. As I said, we have a piering arrangement with the VBNS, so it's entirely possible that a school that is connected -- a university that's connected to Abilene would be in communication with a school that's attached to the VBNS, and so the network, the traffic between the two would go over Abilene and then make a transition at our piering point to the VBNS and vice versa.
JB: That sounds like a lot more institutions than would have access to that high performance networking with that plan.
DVH: Our objective here is to make sure that all of our members have access.
HS: Okay, that sounds good. I think I'd like to turn to applications now, if we could, since we realize that applications are one of the most important things on Internet2.
We do have a couple of questions related to applications, so perhaps we'll start with them. They're from Eric Granstaff, and I'm trying to figure out where he's from. He's from North Central Michigan College. His first question about applications is -- he asks, "What specific forthcoming applications that could be applied in either the traditional or virtual classroom do you see being able to be utilized on Internet2?"
DVH: The most exciting set of applications that are immediately available to the Internet2 user that we think of as being classroom oriented, of course, are the applications that use video and deliver very high-performance, high-definition video services to the user.
For instance, we are working together with our colleagues at member institutions. Just a couple of days ago in Chicago, we launched (together with other partners, including IBM) at Northwestern University -- launched their new digital video activity, and we expect very much that we will, at that facility, be able to store and archive and then deliver anywhere in the Internet2 community very high quality video.
Furthermore, we're working hard to make sure that we can not only deliver archived video, but allow people in one classroom to communicate with people in another classroom using the same facilities, or allow individuals to participate in a class by using video over the Internet. And I'm not talking here about the little tiny square up in the corner of your screen that sort of jerks. I'm talking about quality of video that you frankly don't often get from your local cable television provider.
HS: How is this better than just watching a TV? I know that if my mom looked at this, she'd say, "Big deal! I've seen TV before. You guys are just doing it on a more expensive computer."
DVH: Well, and your mom would be right! It certainly is a more expensive computer.
HS: My mom's right about most things!
DVH: Yes, we've all learned that, haven't we, over our lives?
But the difference here is that you can reach out and get the video you want when you want it. You don't have to sort of look at a program schedule and say, "Ah! I see that program is being shown right now. Now let's see here, I'll have to get my --" Or get your video recorder to record it. In fact, this video is available when you want it.
The second thing, of course, is that this is interactive video. This isn't about just watching. This is about being a source at the same time and talking with other people so that they can see you and you can see them.
JB: What am I going to need actually, then, to do this kind of interactive video? Am I going to be able to do it on just my regular desktop machine?
DVH: As a matter of fact, if your regular desktop machine has a good, speedy network connection and a reasonable video card and a reasonable camera (by reasonable, I mean costing in the range of a few hundred dollars), the answer is absolutely yes.
HS: So that's it? Just an Internet2 connection and some kind of video camera? Is this full-motion video? Is this running MP3 or some other scheme?
DVH: Yeah, we use MPEG.
HS: Okay.
JB: Okay. This is something that people just know that they're going to have a lot of use for in distance learning applications -- and even just our virtual classrooms for our campus instruction, Doug. Is there someone who's piloting this video over the net right now?
DVH: Oh, yes. We have a number of our members who are using the video over the net in various ways today. One of the things we always do is whenever we have one of our member meetings, we set up the production means so that people can watch the member meeting, both in real time in this high quality video, but also the archive is available to them after the fact. So we're using it even for our own business.
HS: Doug, when we talked to you last, you said that this full-motion video could also be used in a multicasting mode. I wonder if you could explain to us the difference between multicasting and broadcasting.
DVH: Well, actually, multicasting is more like broadcasting. And the way video is moved over the network in the production Internet today is what is unicast -- where every individual who's watching a video has an independent connection all the way back to the video server.
Now, you can imagine that, first of all, since the same video is going out of the server to perhaps thousands of people at once, that adds an enormous amount of extra load to the server.
HS: Sure.
DVH: That you don't need. But second of all, that means that -- let's just think about that network link that goes from the server to the first branch point in the network. On that network link, there are literally thousands -- however many people are using the server are coming across that link, each as independent IP streams of information.
Obviously, if you do this right, you could have only one stream of information going to the first branch point and then, at that branch point, send out as many as you have branches, and so on out into the network. And save an enormous amount of capacity in the network and on the servers.
HS: Okay, that means, though, that everybody has to be listening to the same thing at the same time.
DVH: That's correct, and that's what multicast is good for. But it's very interesting. What you discover in many of these situations is at least for a while, people do -- and let's not forget there are other purposes of a multicast. Suppose that what you want to do is update a set of machines with some new software. Well, using multicast, you see, a group of people can dial in, can address that software and the server can sort of wait for a period of four or five seconds before it starts the stream going and accumulate ten or 11 or 100 requests and batch them all up together and, again, save an enormous amount of network capacity.
JB: Let me just jump in and remind our listeners that they can send questions to Doug at expert@cren.net. And we'll be taking more of those soon.
HS: Okay, one question that we have that came to expert@cren.net is another question from Eric Granstaff. And he asks, "Will Internet2 truly be a congestion-free network?"
DVH: No, we don't believe so. In fact, one of the things that distinguishes Internet2 from the Internet we know and have learned to hate from time to time today is that it will operate effectively even when it's congested.
HS: You're going to tell us how it's going to do that, right?
DVH: Right. See, what happened here is, back when the folks who sort of built the Internet protocols thought about all this, they never had a notion that this would become something that would be used by literally tens and hundreds of millions of people across the world. And furthermore, because we were all technical folks, we always thought we'd be able to make the network capacity get as big as it needed to be. So we didn't build the Internet to run efficiently in an environment where some of the links are congested.
And of course, what we've discovered today is that, with the amount of demand and the amount of change in the Internet and demand for Internet services, the fact is that there is always going to be some part of the Internet that is congested. So what we're doing in Internet2 is we're working to build a network that performs even when it's congested.
Now, what do I mean by that? Well, this is what's called our initiative in Quality of Service. The notion behind Quality of Service is that a user of the Internet -- an application that uses the Internet will be able to ask for a certain set of capabilities from the network, and the network will be able to reserve the capacity to deliver those capabilities. And if the network gets too congested, what will happen is some new user will be refused service as opposed to a situation where some existing user who's already working finds out that their service is degraded.
Of course, what this does is takes advantage of the fact that, for a lot of the things we use the Internet, we don't need a really speedy service. We don't need to put the e-mail on the Internet with the same priority that we put a high quality video. And using the fact that the Internet is used simultaneously for so many different applications that have different kinds of priorities, we can uses these Quality of Services approaches to make more efficient use of the network and make sure the applications work even when the network is congested.
HS: How does somebody get high priority in this scheme?
DVH: That's part of what we're working on. In fact, what we have to do is create a set of capabilities around the edge of the network we call bandwidth brokers, that actually have the capability to field requests for high quality service and then examine where the network has the capacity to deliver that. And you heard last week from Ted Hanss about middleware. Well, in fact, a very important middleware application is the middleware that allows us to manage these bandwidth brokers and serve the users of the network. We're just beginning the development of those bandwidth brokers.
HS: Are people going to have to pay more to get higher priority?
DVH: I think that that's one way it will be done. I think that, you know, different --
HS: Actually, I don't see another way that it would work. Otherwise, everybody, I think, would want to have highest priority.
DVH: Well, certainly it's the case that the market mechanism is one of the most efficient rationing mechanisms we know about, and I expect it will be the dominant one that's used here. But there are many other ways to allocate capacity than just simply to ask people to pay more.
HS: Okay, one last question from Eric Granstaff -- and it's quite a change of pace from what we've been talking about, but it's an interesting question. Eric asks, "What market changes are in store for telcos and other emerging telecommunication companies?"
DVH: I think that's a really interesting question, and I'm absolutely certain that when I answer the question, I'm going to be proven wrong sometime in the next five years, because I think we can't really imagine --
HS: Five years is okay, as long as it's not five minutes!
JB: Five years is a long time!
DVH: Five years is an eternity on the Internet, isn't it?
JB: How about the next five weeks or five months?
DVH: The first thing that I think is going to happen is that the Internet will, over the next several years, become the carrier of choice for all of the kinds of media that we now think of as separate. We're going to see TV on the Internet; we're going to see all of our telephony on the Internet; we're going to get our videos over the Internet. We're already starting to get our CDs over the Internet.
The result of that is that telecommunications is going to shift away from providing a set of independent services -- whether it's cable television or telephones or data networking -- into an environment where all of those services are provided through one set of communications protocols.
And as that happens, large telecommunications companies that have specialized in one part of this space and built their infrastructures to accommodate that one part most efficiently are going to find that they have to rebuild their infrastructures to handle this more general kind of traffic.
Of course, we can already see that going on. A large number of the major telecommunications companies have already announced that they're moving into this kind of a future and I expect we will see more doing that.
HS: Okay. Doug, now that Abilene (at least probably from your point of view) is behind us -- there are a lot of people who aren't on it yet. That's probably not their view. But from your point of view, now that it's behind us, what do you see as the next big challenge?
DVH: I wouldn't precisely say Abilene is behind us, but Abilene is working. And so now the question is, what isn't working that we need to get working, right?
HS: Yeah.
DVH: And as we travel around the country and talk to the faculty at our member institutions, what we hear more and more is that, despite the fact that we have Abilene and the VBNS providing backbone network services, and despite the fact that the university is connected with a high bandwidth connection, those faculty are having a hard time getting access to this service on their desktop. So we've got now a challenge of making sure that this capability reaches out into all of our campuses. That's the first challenge we've got.
The second challenge we have is that of course the reason these faculty members want to use this is they want to do new applications. They want to use these new capabilities we've been talking about and in many cases, they haven't been able to find the support and the knowledge they need to do the applications development and use the new network capability. So our next focus is going to be moving forward to make sure that we enable our member universities to deliver the capabilities to the faculty that are required for us to move forward here. And that's going to be a complex process that will require more cooperation in the networking arena than higher education has seen in the past.
HS: And you think that Internet2 is going to actually help that along? I mean, folks cooperate kind of when they have to.
DVH: Well, folks cooperate when there's something to be gained from cooperation, and Internet2 provides an exciting set of potentials that, if we can cooperate and work together, will be available to all of us.
HS: Okay. We just got a question from Laurentian University from Richard Danielson, and he's wondering about the difference between Internet2 and an ATM network.
DVH: Oh, that's a very good question. In fact, ATM, as he probably already knows, is a particular low-level protocol for carrying data in a network, and in the very high speed backbone servers -- VBNS, for instance -- ATM is one of the protocols that's used to carry the data through that network.
In fact, that network uses the SONET protocols at the lowest level to put the bits over the fiber, and then it uses ATM as the next layer to break the bits up and allocate the capacity, and then it uses the IP protocols (or what we call the Internet Protocols) at the next layer. It's not necessary to have ATM in an Internet network, but it in many cases adds value because it allows you to allocate the capacity of the network in ways that today it's difficult to do with just the Internet Protocols.
JB: Is that how the Abilene network is structured?
DVH: No, the Abilene network doesn't use ATM. The Abilene network uses SONET and then it uses the Internet Protocols without using ATM in the middle. And one of the reasons we chose to do that is because the folks over at the VBNS at MCI Worldcom were already doing it with ATM, so we thought it was good to have a backbone that operated without ATM so we can begin to learn about the strengths and weaknesses of the various approaches.
HS: Okay, so ATM is just one of the many ways that one could implement the features of Internet2.
DVH: It's one of the tools that we have available to us, yes.
HS: Okay, in my intro, I mentioned that one of the surprising applications on Internet2 was middleware. I wonder if you could talk a little bit about the middleware or middleware-like applications on Internet2?
DVH: Well, certainly. I was talking earlier about this challenge of making these applications work in the Internet2 environment. Earlier on, we were talking about the notion of a classroom where we were delivering video. Now, suppose that what you wanted to do in your classroom was do something that's similar to what we're doing here today but with video, so that in the middle of a class, you called up a colleague and had that colleague deliver a mini-lecture in the middle of the class. Well, to do all of that with today's Internet technology is a very complicated process, and once you get it done, it's not clear that it's going to be reliable.
What middleware really allows us to do is to begin to make those kinds of applications routine. One of the objectives we have is to get to the point where, within 15 minutes, any faculty member at any of our member universities can be in a video conference with any other faculty member in the Internet2 community. That requires a whole set of software tools that allow those people to find one another, that allocate space for them on the network, that allocate a place for them to meet in the videoconference, and those tools -- while they exist in various sub-communities of our community -- are not consistently available across the Internet2 university community today.
And until we do get that consistent layer of software in place, we're not going to be able to provide a convenient access to many of these exciting applications that Internet2 provides.
JB: We have a related question coming in, Doug, from John Collins at Duke.
DVH: Yes.
JB: And he's asking if you expect the Quality of Service for differentiated services to be done with IP as a bearer service or with tools like ATM or in ways that we don't know yet.
DVH: We expect that the Quality of Service will be done in many different ways.
One of the things we very much hope we can work to arrange is to have various network clouds use different detailed technical schemes to deliver Quality of Service -- but define an interface at the boundaries of those network clouds so that one network cloud that, say, uses ATM and another one that uses IP as the bearer service can in fact provide Quality of Service across the boundaries between those networks.
We don't want to try to dictate how people accomplish Quality of Service, but we do want to work very hard to make sure that Quality of Service works across many networks simultaneously because, of course, in higher education, almost all of the things you do uses more than one network between your computer and the server, or your computer and the person you're working with.
JB: And I think a second part of that question is, how long do you expect this development to take before widespread deployment? I think another way of asking that question, you know, is how soon does one start planning programs based on this type of Quality of Service?
DVH: We have right now an initiative under way we call the QBone initiative --
JB: Excuse me, is that Q and then Bone?
DVH: Yes, that means Quality-of-service Backbone, right?
JB: Okay.
DVH: QBone! And the QBone Initiative -- actually, a segment of that initiative is a number of universities and research labs actually all across the world beginning to experiment with these Quality of Service protocols to see what works and what doesn't work. We hope that sometime later in '99, before the year 2000, we will know a lot about what works and what doesn't work, and we'll start being able to do trial implementations of the use of Quality of Service in the Internet backbones and the networks on the campuses and at the Internet GigaPops.
The interesting question about production is not so much when will we be able to actually allocate the capacity of the Internet (that is a big issue and I expect it will be. We're hoping to be in pretty good shape on that within a year or so); the question goes back to this middleware question we were talking about a little bit before. The question is, if you're sitting at your computer at a member university and you need a certain Quality of Service, how long will it be before we have the middleware that will allow you to get that flexibly and automatically from the network? And that is going to be an even bigger challenge than actually making sure the bits flow through the network correctly.
HS: So you imagine that I could go -- when this thing happens, when this middleware is in place -- that I could go to any university that's a member and I could be authenticated, authorized, use directory services on any other member system. And that would work.
DVH: Yes. That's a very important part of this. We would very much like it to be the case that students and faculty at any of our universities can be at another university and appear to be virtually back at their university.
HS: And how far along is that? I mean, it sounds like that's going to take quite a bit to do.
DVH: It's going to be a very tough task and it will, of course, depend on how fast -- not only how fast we come together and decide what the appropriate standards are for making that possible, but also how aggressively the individual universities implement those standards.
And of course, it will be like everything else in higher education. Some universities will do this before others. We very much hope that, as a result of our Middleware Initiative, within a year or so we'll have at least a small number of our members beginning to implement those capabilities.
And I very much hope that we can really make that a reality for most of the higher education Internet2 membership within a couple of years. I've got friends that tell me that I'm hoping for too much, but that's what I hope for.
JB: Given the way 'Net time happens, perhaps we can afford to be optimistic.
DVH: Well, I'm certainly hoping that's true,�
JB: We've been talking a lot about video and Quality of Service and all that. Maybe we should move on to a couple of our questions having to do with Internet2 applications that are helping researchers do the kinds of things that they want to do. Is there an example there of an Internet2 application that's really making a difference right now in how research is being conducted?
DVH: I think that some of the most exciting applications we have today in the Internet2 institutions are those applications where researchers are collaboratively using experimental facilities and remote instruments.
At the most sort of straightforward level, we now have at a number of our Internet2 institutions microscopes online which are set up on the Internet so that people at a remote location can send a specimen in and view it with the microscope without having to be where the microscope is.
A slight extension of that is when there are several people located around the country who have a common interest in a specimen, but the specimen may well be a specimen that will in fact (once you put it on the microscope) begin to deteriorate. So what you want to do in that case is everybody wants a chance to sort of participate in that viewing of the specimen, and to talk with one another about what they see as that happens. That's difficult to do even if everybody could get together in the same room, but if you equip a microscope with the necessary Internet technology, they can all be in their offices and watch that viewing of the specimen. And we have that application beginning to happen.
HS: It sounds like, watching it you could record that and play it back later. But it sounds like what you really want is people to be also manipulating.
DVH: That's the key. You're absolutely right. The key is that as things go along, you want to look at different parts of the specimen. You want to change the lighting on the specimen. There's things you want to do and you want to be able to talk about what you're doing so you don't miss something important.
HS: So it's this whole issue of remote instrument control.
DVH: It is.
HS: And you're talking actually about remote instrument control being done by many people simultaneously.
DVH: Yeah, although obviously one of them has to have the controls. But they all get to discuss how this is going to happen.
HS: Right, they could hand the controls off to people, one person to another.
DVH: That's exactly right. You've got it.
JB: Sounds to me that's a little bit like who has the remote?
DVH: There you go!
HS: Well, we know that in lots of places, there's more than one remote.
JB: That's true. Is there another application in a particular discipline group, Doug, that has a project going on that you wanted to mention?
DVH: Well, we've already talked about getting people together in virtual meetings using video. That's a very important application for the high energy physics community because their big instruments are at places like Serne in Switzerland and Slack in California and Daisy in Germany and they can't all be there. They have to be on their own home campuses, but if they want to participate in the leading edge of discovery in high energy physics, they need to be able to talk with the physicists who are there.
But perhaps even more important, they need to have access to the data that's coming out of those experiments, and they need to be able to analyze that data themselves. And so a very important initiative we have under way now is to make sure that we help that high energy physics community get the connectivity and the capability they need so that, for instance, physicists here in the United States will be able to play a major role in the experiments going on in Switzerland at Serne.
HS: One of the applications you mentioned before, Doug, that doesn't sound much like an Internet2 application, is data mining. Could you say a few words about why data mining is an interesting application for Internet2?
DVH: I certainly can. First of all, let's talk a little bit about what data mining is. You know, all of us have become familiar with the power of computers to do data analysis, but suppose, in fact, the data that you wanted to analyze was spread around a half dozen or a dozen different places in the world. And to do the data analysis you want to do, you need to have all that data in the same data set.
Well, you can't just move all that data to a single place. What you have to do is set up a means of dealing with it as if it were in a single place, using the power of the Internet. And you can imagine how much -- especially for very large data sets -- how much network capacity it takes to bring all that data logically together so that it could be analyzed. And in some cases, analyzed using parallel computing systems so that you can in fact get through these very large data sets and do interactive analysis of them. We've discovered in our trials of the data mining applications that those folks are able to saturate almost any network we can put up.
HS: So this looks like distributed disk storage that's distributed across the network.
DVH: It's actually a little more complicated than disk storage because each of these data structures, of course, is different, and you need a meta definition language so you can actually match them up. So transformations need to actually happen in the data on the fly.
JB: That sounds like more middleware is needed to do that. Is that true?
DVH: Well, in fact, this software that allows you to actually describe these data sets using a common meta language is right on the boundary between middleware and applications software.
HS: In the beginning of the broadcast, Judith said something about this meeting that's coming up in the spring, the Internet2 meeting.
DVH: Yes.
HS: Could you tell us something about what's going to happen out there?
DVH: Well, I happen to be in Washington today. It just happens to be where I'm located and we're going to have our spring meeting of the Internet2 membership here in Washington later this month. And we already have a very large contingent of folks coming that the meeting will be held on Wednesday the 28th here in Washington. And the reason we know that there's a very large contingent of folks coming is because the hotel room blocks that we reserved are already sold out.
And of course, at that meeting, we'll not only discuss many of the kinds of things we've been discussing in this call, but we'll be talking about the next steps we need to take to move the Internet2 effort forward. We'll be showing demonstrations of many of the kinds of applications we've been talking about today, and furthermore, we'll be making sure that we understand (since we're having the meeting here in Washington) how the Internet2 activity relates to the activities of the next generation Internet program and the federal government and the federal government's new Information Technology for the 21st century -- or IT2 initiative that is just beginning and is proposed here in Washington. So we expect a very exciting day-and-a -half here as we move forward into the next year of the Internet2 effort.
JB: I can't believe our time is getting so close to the end here. I'd just like to ask our listeners, this is kind of a last call for some questions for Doug. And Howard, you had another question?
HS: I have lots of questions here, Doug. If you want to hang around here until late in the evening, we could continue.
DVH: I might get hungry.
JB: Well, we haven't figured out how to do the pizza over the 'Net yet.
HS: That would be an interesting Internet2 application, and we'll talk about that later here. But how's the membership growing in Internet2?
DVH: We're now at about 150 universities as members of Internet2, and there are now 50 corporations that are members of the Internet2 effort. We don't expect there to be really substantial growth beyond this level because, remember, Internet2 is really an effort that is focused on pushing the frontiers of the Internet forward, and that is an activity that demands a lot of resource and is not the kind of thing that every university wants to get involved in. So of course, we welcome all the universities in the United States that really want to be partners in advancing the Internet and Internet technology and we'll continue to have some growth. I wouldn't be surprised if we maybe get up to 170 or so, but I don't expect that the Internet2 university membership will grow a lot over the next couple of years.
HS: The reason I really asked that question was, you described all kinds of really interesting things that people are going to be able to do and I was concerned, when you said the membership was small and not expected to grow much. In fact, that's the way you wanted it. That means that only these 150 people, right, will be able to participate in all this neat stuff?
DVH: Well, if you've got 150 universities, you've got more than 150 people!
HS: Sure, I understand you have more than 150 people. I'm sorry, I meant 150 --
DVH: Yeah, you've got probably over a million people. But your point is very well taken, and that's why we're working so closely with our corporate members because, you know, as we develop and test this technology and get the flakiness out of it and so forth, then what they're going to do is they're going to incorporate it in their products and their services so that people will be able to buy these capabilities directly through the commercial marketplace.
Furthermore, we're working closely with our GigaPoPs, most of which are part of regional networking organizations in various places around the country, and those GigaPoPs are also beginning to think about how they will deploy these capabilities within their region to support the members of their regional networking organizations.
So this technology, even though we won't expand the membership of the Internet2 project itself rapidly, the technology -- we're working very hard to make sure that that doesn't just leak but actually we're designing it so that it migrates in a smooth way out into the broader Internet.
HS: So when you say "in a smooth way" you mean that bit by bit, it's going to come out? It's not going to come out all at once?
DVH: Oh, absolutely not! In fact, we're already beginning to see little pieces of it show up.
HS: Such as?
DVH: Yeah, you know, both MCI Worldcom and Quest have announced 2.4 gigabit Internet service availability, and that was something that only Internet2 did when we started out.
HS: Is there anything that regular folks (when I say "regular folks," I mean, I'm sorry, you regular folks) anything that non-Internet2 members can do -- should be doing to either keep track of this or to get ready to participate?
DVH: Well, the first thing I have to say, I'll put in a plug for programs like this. I mean, we can all stay up to date and CREN has been doing a great job and Educause is doing a similar high quality job of sort of making sure that -- we're working closely with CREN and Educause to make sure that this information about what we're doing and how it's going and so on is available to the whole community, not just to our members. And that allows people to plan and make the kinds of decisions they need to make so they'll be able to participate.
The second thing I think that is important to do is, if you're a higher education institution, to get involved in your regional networking activity and get connected and get your technical people staying up to date on what's happening there because my expectation is that those regional networks will in many cases be the places where you'll get earliest access to these new capabilities.
JB: I think that's really very good and practical advice, Doug. I think maybe we want to think about a final question. You know, one of the things I remember your saying that really struck me is the fact that the goal of the Internet2 applications is to be something that we all do routinely, just regularly, as part of how we work. Do you personally use the capability of the Internet2 right now? Did we agree that I could ask you that?
DVH: That's a fine question, and the answer is at the place where I routinely work today, I don't have access yet to the capabilities of the Internet2 backbone. We just installed the conduit to bring the high capacity into our Ann Arbor offices last week, and so in another couple of months (I hope sooner than that) I'll have access.
But of course, I do visit member campuses, right, and I get to use these new capabilities when I'm visiting those campuses. And I'll give you a very mundane, very mundane example. I was at the University of Wisconsin in Milwaukee the other day and my computer was there with me after I made a presentation. I was sitting in the audience, and of course, I was interested in what was going on. But it struck me that I hadn't backed my hard disk up recently on my computer, and there was a 100 megabit Ethernet connection with a path through Internet2 available right next to the chair I was sitting in, so I just attached my laptop to the 100 megabit Ethernet with 100 megabit Ethernet PCMCA card in my laptop and I started the backup. And it did the full backup of my hard disk in about ten minutes. And the data rate that I saw moving across the network was a little over 12,000,000 bits a second.
JB: Oh, wow. It sounds like a small thing, but I bet there's a lot of folks out there that say, "Yes, I'm ready! Any time!"
DVH: Well, it's not a very exciting application, but the fact it that I feel much safer when my computer has got a backup located in a secure machine room 1,000 miles away.
HS: Yeah, rather than sitting on the diskettes or the zip drive on your desk.
DVH: Right, exactly.
JB: Okay, Howard, do you have a final question here before we wrap up?
HS: Well, there's lots and lots of questions here, but I think that any one I could ask is just going to start us into something. So there's nothing short and simple, I think, that we could do here. But it certainly has been very interesting, and for the folks who did send questions in (there were a couple sitting out there that we did not get to answer), we will have Doug answer them and put them in the archives.
JB: All right, well, thank you very much. And at this point, let me say thanks to all of our Web participants for being with us today for this time with Doug.
And be sure to mark your calendars for two weeks from today, which is April 22, and the TechTalk will feature Frank Drew from the University of Minnesota and Jeff Hodges from Stanford University and they'll be talking about campus directories. One of them, Stanford, I believe, is using LDAP and Frank is using something else. So do plan on joining this session and inviting your friends and colleagues.
DVH: Let me point out, that campus directory issue is one of the key middleware issues that we're going to have to work on.
JB: Good. Thanks for mentioning that, Doug. I know from what I've seen, too, it's a two to three year project. It's not something you can decide to do and just do, so it's something important to be planning for that. And I'm sure we'll hear more from the folks on that.
So thanks to everyone who made this possible today: the board of CREN; you, Doug, and technology anchor, Howard, Web content producer, Terry Calhoun; and to Harold Ansell and Lee Perlis of CREN; Paul Bennett and Martha Vander Kolk at UM Web services; and all of you for being here. You were here because it's time.
Bye, Doug. Bye, Howard. Bye, all.
HS: Goodbye, Judith, goodbye, Doug, and goodbye, my friend zebra.
JB: Zebra! We don't let them forget.
