TechTalks Transcript
What's New with Internet2?
 Judith Boettcher [JB] |
 Howard Strauss [HS] |
 Ted Hanss [TH] |
Nov. 19, 1998
Audio
• Streaming
MP3
• Download
MP3 (Download
Tips)
JB: Welcome to the CREN TechTalk series for Fall of 1998, and to this session on "What's new with Internet2?" 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 the technology anchor for TechTalk, Howard Strauss from Princeton. Howard is a well-know Web and all-around Information Technology expert.
Welcome, Howard.
HS: Thank you, Judith, and thanks again to CREN for choosing me as the technology anchor for TechTalk.
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, Ted Hanss, 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 archive later.
Today, most people probably take the Internet for granted -- it often seems like it was always there, wasn't it? And we just assume it will be as reliable as the electricity delivered by our power companies. But it is not. Unlike our electric wires, our Internet connections get clogged with packets. And sometimes they're simply not available at all. We end up waiting endlessly for some Webpage or other to appear on our screens.
How nice it would be, we have all thought, if the Internet were a zillion times faster than it is now, and lots more reliable -- if our Webpages, e-mail and large files were always delivered instantaneously, without delay.
The answer to these needs might be in a project called Internet2, which promises to provide us with those faster, more reliable connections we're all waiting for. But what else will Internet2 do for us? Will you and I be able to afford it? And more important, how and when can we start using it? (Because I would sure like to do that!) Our guest expert, Ted Hanss, will try to answer these questions and many more, on today's Webcast of TechTalk.
Judith?
JB: Thank you very much, Howard. And let me give our guests just a little more detail about Ted. Ted is the Director of Applications Development for Internet2, and as most of you know, Internet2 is a project under the direction of the University Corporation for Advanced Internet Development, often--sometimes referred to as UCAID. (And we'll have Ted tell us much more about all of these things very soon.) Ted hails from the University of Michigan, where he was a senior manager in the University of Michigan's Information Technology Division before joining Internet2. I'm also pleased to say that Ted was one of the key presenters in the latest of a series of CREN Virtual Seminars on Internet2, jointly funded by UCAID. This new seminar is now available both on the Web and on CD.
Welcome, Ted, to TechTalk and thanks for joining us here.
TH: Thanks, Judith. Thanks, Howard. I'm glad to be here.
JB: All right, great. Well, let's give our Webcast perhaps just a quick roadmap as to where we might go in today's session. We've got three different sections today, really, and the first one's starting out and launching with some basic information about Internet2. And then we'll talk about, perhaps, the structure of Internet2 and how people can get connected to Internet2. And then we'll move on to Ted's real area of specialty, which is examples of applications running on Internet2.
HS: Actually, I think that all those areas are Ted's areas of expertise, and I think he'll agree.
JB: Well, that's true. That's true. It just slipped a little bit there, Ted.
HS: Ted, I wonder if you could just start out by explaining how Internet2 is different from the thing we have now, just to set some groundwork for this thing that perhaps people are going to be using in the future.
TH: Well, one of the things that's different about Internet2 is that it's an environment for our university members to engage in applications development, having access to resources that aren't available across the commodity Internet. And essentially --
HS: Is that what we're going to call Internet1, or whatever it is? The commodity Internet?
TH: Yeah, that's right. Yes, that's sort of the colloquialism for what you can, you know, buy off from any Internet Service Provider we call the commodity Internet. It's a commodity service now. You basically get the same function from everybody, and what happened was, those everybodies out there turned out to be tens of millions of people accessing Internet services. And they're using up the capacity in a way that some of the really exciting applications that the faculty and researchers in our campus want to develop don't have the network capacity in order to do those kind of explorations.
So what's different about Internet2 is in the first case, it's basically Internet technology that's exclusively for the use of the member universities -- much higher bandwidth, and at the same time, we are trying to innovate on some new features.
HS: So it's not going to replace the commodity Internet?
TH: No. Unlike the experience we had with the National Science Foundation Network, NSFNet, back in 1995, we don't see a process of privatizing. Instead, we see experiences in the applications we develop through Internet2 being transferred out to the broader Internet and raising the capability of that overall.
HS: But could you talk a little bit about, I mean, how is it going to be different? Is it just going to be faster than the commodity Internet? Anything else going to be different about it?
TH: Well, faster for one. We're putting in the very leading-edge technology to provide very high-capacity network links among our universities.
But that's not enough. We're also looking at ways of increasing the functionality of the network. And examples of the include, we're putting in Quality-of-Service features. And what that means is that it allows an application to get higher priority on the network so that, for example, video-based applications will be delivered more quickly and reliably, while things like e-mail and file transfer (that can occur more slowly) can happen in the background. That's one big difference.
The second thing we're working on is a technology called multicast, which makes more efficient use of the network when you're delivering lots of information to lots of people. So instead of sending the same copy seven times to seven different people, it'll send one copy out and then will be duplicated at the client end, as opposed to from the server.
JB: Why don't we follow up for just a moment on that multicast idea, Ted. Is that anything like broadcast? How will people use this multicast technology?
TH: Well, it's similar to broadcast, but what it is, it's only directed to those people who want to subscribe to a service. Right now, for example, television stations send out a broadcast, as we define it, that goes out to every house. Even if you don't have your TV on, your antenna's capable of picking it up.
With multicast, it is a service that only those who subscribe to it actually receive it. So if you want to get a video sent to you via multicast over the Internet, you can get it at your desktop while the person at the next desk doesn't have to get it.
HS: But does that mean that if a hundred people want to receive a multicast thing, that there are a hundred copies of the packets sent out?
TH: No, that's exactly what we're trying to avoid.
HS: Then how do we avoid that, when a hundred people have to receive the packets?
TH: What happens is, there's intelligence in the network so that the server sends out one copy, but as it goes to the Local Area Network where the clients are, if there's more than one client there, it will deliver it to that more-than-one-client, but it won't make 100 different streams taking place.
HS: So we really need some complete network structure to do this Internet2. This is not going to use the same hardware at all that the commodity Internet uses. Is that correct?
TH: It's going to be very similar hardware, but what it will be -- there's some upgrades in the software that runs, for example, in the routers (the boxes that direct the traffic) and there is to be a lot of experimentation. We're going to learn about what it means to provide those upgrades on the routers. And based on our experiences, then other Internet Service Providers will be able to make those same kind of upgrades and that functionality will be made available more broadly beyond just the university community.
HS: You keep talking about the Internet2 being used by the university community. Does that mean that if you're a corporation or an individual or if you're not part of the university -- forget it, you can't use this Internet2 stuff?
TH: Well, the Internet2 project's been around for almost two years now, and the membership has been open to universities to participate on basically a self-selecting basis. That is, any university that supports our mission of developing advanced applications can join the project.
Now, what's happened is that we are providing network capability called the Abilene Network that's going to join the NSF and MCI partnership of the Very-high-speed Backbone Network Service, or VBNS. And at the same time this fall, both the National Science Foundation and Internet2 have opened up participation and access to the physical network to corporate research labs.
So we're now just -- we have not anybody connected yet, but you can imagine organizations like Bell Labs or IBM research or even groups like Ford Motor Research who are doing research and development with the universities connecting into the physical network and participating.
HS: Could you talk a little bit about how the thing's being paid for? It sounds like quite an expensive kind of thing.
TH: Well, there's funds coming from different ways to pay for this. One part, each organization that joins Internet2 pays an annual membership fee in order to --
HS: How much is that? Can we talk about prices over this Webcast?
JB: Oh, sure. You certainly can.
TH: For university members, it $25,000 a year, and --
HS: That's independent of the size of the university?
TH: Independent of the size of the university. And the majority of our members -- we have 135 university members now -- the majority of them are the Research I institutions, so most of them are pretty large-scale institutions. And they pay $25,000 a year which goes and pays the salary of people in the central organization. It goes and pays for the meetings (you don't pay extra fee to come to one of our meetings). And it also supports the various activities we have under way in terms of work groups and so on.
But that doesn't pay for any physical networking of the structure. Each university, when they join Internet2, also makes a commitment to upgrade their campus infrastructure to provide high-capacity links down to the desktops and into the laboratories and so on. And that kind of investment can be anywhere from half a million dollars a year to over a million dollars a year per institution.
And then there's the regional and national networking links. And those are quite significant. Those go up into the tens of millions of dollars costs, and the National Science Foundation provides some subsidy for those links through their High Performance Connections Program. In addition, we have each university contribute probably the major portion of the cost for the connection. And thirdly, we have donations from corporate sponsors who -- for example, some of the networking vendors like Cisco and Fore and 3Com have contributed hardware that will allow universities to connect. And then the Abilene Network and VBNS, MCI and Quest and so forth, have contributed capacity to these resources.
So it really is a government, university and industry partnership that's providing all the support for this.
HS: You said, Ted, that there's 135 members right now. There's obviously lots of universities out there -- there's a lot more than 135 universities out there -- lots of universities out there that are not participating. If there's some university out there that wants to participate, what are the first few steps they have to take, and what's their kind of financial commitment? How much do they really have to do to become part of this?
TH: Well, the first thing they do is they contact us for a membership application, and you can go to the Internet2 Website (which is listed on the CREN Web page, but it's also www.internet2.edu). And you can find the place where you can request membership.
And essentially, what happens is that the executive in your institution (for example, the president or the chancellor) will make a commitment on behalf of the institution to support the objectives of the project which -- as I say, for the university, everyone pays a membership fee and then you also make the commitment to support the network infrastructure, which you may or may not have already committed to. And so again, those kinds of costs vary. But right now, we just basically accept those membership applications from any university that can make that commitment.
There are other universities who aren't ready yet, and many of them prefer to follow along by looking at the Webpages and see what information we're sharing. Or they may work in a regional basis with other universities -- research universities in their area, for example -- and partner on activities. Or they themselves may decide that the best way to stay in touch is by going to conferences like Educause and CUMREC and others, where we share information about our projects to the broader higher ed community.
JB: Ted, what about the timeline? I think we've been looking at the development of Internet2. Is there any part of Internet2 that's actually up and active right now?
TH: Well, of our 135 member universities, a majority of them have written for grants from the National Science Foundation for these high performance grants. And right now, somewhere around 70 of them -- just about half of them -- actually do have a high performance connection to the Very High Performance Backbone Network Service, or VBNS.
HS: So the other half are not really doing anything (inaudible)?
TH: The other half are basically -- the time frame for getting these circuits is fairly lengthy. And so even if they've received the grant, it still can take weeks and months to actually get the high speed circuits in place. So we actually have progressed pretty far in the last couple of years, where we went from just a handful of universities to what we have now. And next year, we'll certainly have well over a hundred universities connected either to the VBNS or to the Abilene network.
HS: Could you tell us the difference between those? You keep mentioning VBNS, which I think is an MCI thing, and Abilene.
TH: Right. As you said, VBNS is a cooperative agreement between MCI and the National Science Foundation that runs from 1995 to 2000, and recently MCI also announced a commercial service based on that technology which they'll continue offering beyond the 2000 end of the agreement with National Science Foundation.
The Abilene network is one that is actually a project of the University Corporation for Advanced Internet Development with partnerships with Quest providing the fiber capacity, Northern Telecom providing the equipment to light the fiber, Cisco to provide the equipment to move the packets around, and Indiana University providing the network operations center.
Now, there's two networks and some of our members -- you know, basically to an end-user on campus, it's invisible in terms of what network you use.
HS: Then why are people being urged to switch from VBNS to Abilene?
TH: No, they're not really being urged to switch. What we're basically offering is options for universities that depend on where you are physically in the country in terms of the connections you have. And in some cases, there are some actual differences. For example -- and the end result will be, some universities will be on the VBNS, some will be on Abilene, and some will be on both.
But because we'll allow information to flow from one to the other, and we have an agreement with MCI to do that -- MCI and NSF -- to the end-user on the campus it won't matter if you're at a university on the east coast and there's a university on the west coast you want to get to. It won't matter which backbone you're on.
JB: So it will be somewhat seamless, like the operation of the current network is?
TH: Right. And that's exactly the idea: that it's for the higher ed members in the Internet2 environment, that it is seamless.
HS: But I thought that Abilene either had more bandwidth or was faster or was more reliable -- that there were some real advantages to Abilene. Did I miss that?
TH: Well, there's a couple of things there -- features that are distinct between the two of them. For example, the VBNS network uses a technology called ATM, or Asynchronous Transfer Mode, which is a technology that allows you (in sort of the simplest of terms) to dedicate a link between two end-points for a certain application to get the capacity that it needs in a way that if you want to insure that you get 4Mbps between any two end points, this can reserve that.
Now, in the Abilene network, we actually are not using ATM, and we're trying to figure out ways that we can use the TCP/IP protocol to do the same sort of thing. But we're sure that IP protocol is further behind ATM on that. The other thing is that right now, Abilene will start off being a faster network in terms of our goals of being at 2.4 gigabits-per-second next year, while the VBNS right now is at 622 megabits-per-second. But the VBNS folks have also made a commitment to upgrade to 2.4 gigabits per second, so we're also increasing the capacity of both the networks.
HS: You imagine in the long term they're going to both coexist?
TH: Well, yeah, the idea is for them to coexist. I mean, the only distinction may be that there are more access points for Abilene around the country than VBNS. And so some universities may find it's less -- it's more affordable for them to connect to Abilene because there's a connection point nearer their university. But for some cases, the opposite may also be true.
So essentially, a university has to go through a decision point. -- Is ATM important? Where are the connections, the points of presence and so forth? -- before they made the commitment to which of the two, or both, they connect to.
HS: Okay, one of the things we talked about a little bit earlier in the beginning was that one of the reasons for doing the whole Internet2 thing was to put up some advanced applications (and I lead an advanced applications group, so that has a lot of interest to me). I wonder if you could tell me what kind of advanced applications are we talking about that might run on Internet2 that couldn't run effectively on the commodity Internet?
TH: Right. That's basically the ones we're targeting are those applications that support research and education and don't run on the current commodity Internet. And while the history of Internet technology in higher education had been focused on linking universities to supercomputer centers, our focus on Internet2 is every discipline across the campus -- whether it's science and engineering or arts and humanities.
Now, examples of the applications we're using: one of the biggest area's we're trying to promote is collaboration environments -- allowing people to work with each other as if they were sitting side-by-side but they are in fact separated by time and distance. And some of the things that augment that are delivering high-quality audio and video across the network. Not the postage stamp video that you get today over the commodity Internet, but full-screen, full-motion video.
JB: So -- I'm sorry, go ahead, Ted.
TH: I was just going to say that that's not just for people to sort of look at each other when they're talking. It's also so we can do things like drive remote telescopes and remote microscopes across the network -- you know, very expensive resources that you don't necessarily have on your own campus or within your own lab, but you can get to them over the network using that kind of --
HS: Is anybody doing that? That sounds like an incredible kind of thing -- to be able to sit in my office and operate some telescope halfway around the world.
TH: Telescopes are the hardest ones, actually, because the telescopes themselves are usually on top of a mountain, so not near high-performance connections. But that's being worked on.
More prevalent are the remote microscopes, such as scanning electron microscopes at very high resolution look at objects, whether it's chips that are being developed or you may be looking at microscopes that go after biological samples. And it is something that right now is more and more increasingly used -- is doing that from hundreds of miles away. You know, you FedEx your sample to some location, they put it in the microscope and let you view it at your desktop without having to have traveled to the remote location.
HS: But is that something that's just planned, or is there actually somebody doing that kind of stuff?
TH: Actually occurring right now. There are efforts everywhere from, say, the University of Minnesota to the University of California San Diego to the University of Michigan and many others doing this kind of applications in production today.
JB: Do you have examples of these kinds of applications on the Internet2 site anywhere, Ted?
TH: Yes, we do. Up on the Internet2 website there are lists of example applications.
HS: One of the applications that you mentioned in one of the times I talked to you that I was surprised, actually, to hear about was you were talking about middleware. I think of middleware as something that MIS folks have to spend their time doing, and I was surprised to hear that as being an advanced application on Internet2. Maybe you could tell me something about middleware and Internet2.
TH: Sure. What middleware is are the common software components and services that most applications need. And the kind of things we talk about, for example, are security services or authenticating and authorizing access to content, whether it's library content or remote microscopes or directory services to try to navigate and find resources or things like the technology for encoding video.
And what we've found is a lot of the -- as is obvious, the application developers on our campuses are faculty who are the chemists or the biologists or the professors of physics who don't want to have to deal with all those other issues of security and so on. And they want to focus on the science aspects of their application.
And so we've launched an initiative to try and provide a base set of services that can be made available to application developers so that each developer at each university doesn't have to duplicate the work at each of their -- each time they develop an application. It's essentially taking some of the experiences we've had, as I said, in the MIS areas, but now applying it to a wide area Internet applications.
JB: Ted, you mentioned when we were talking earlier about that, there were two important projects that got announced at the Internet2 member meeting in September. Did you want to maybe talk a little bit about those two projects?
TH: Sure. The two projects that we announced as part of the new Internet2 middleware initiative are the Internet2 Digital Video Network and the Internet2 Distributed Storage Initiative.
The digital video network is a project that is uniting all the various digital video projects occurring at our universities into a comprehensive effort so that we can all learn from the experiences and practices developed at other universities. Everything from learning how you can go and set up an on-demand video server to how you do gateways for video that goes across one type of network to another type of network. And we're learning a lot from each other by moving forward and sharing our experiences. We basically want to make video as easy to use as it is to send around text attachments on e-mail today.
HS: You mentioned the other project, this distributed storage initiative. Let's talk about that for a bit, could you?
TH: Sure, in that project, the idea is to look at how we can -- even though we've got high-capacity networks, we still have very, very large data sets around, whether it's video data or information coming from high-energy physicists who work at Sern over in Switzerland and gather instrument data from there. And so we're trying to figure out how we can intelligently distribute that data on the network -- whether it's Webcasting or other kinds of replication and distribution modes -- so we can get data to the scientists who need it.
HS: Could you give me an example of how that would work? I'm a scientist and I'm running this cyclotron or whatever they have out in Sern.
TH: Well, for example, they have some new experiments coming out on line the next decade that are going to be producing terabytes of data a week, and there'll be regional data centers around the world, and there'll be one in North America. And so one thing we have to think about is how do we easily move that data from Europe to the North American site. And then there may be a desire within North America to replicate portions of the data set out to locations closer to some of the universities, so that they don't have to go across the backbone network to get it. And instead of just having it be in sort of passive caching, we may want to proactively distribute it out to sites around the network.
So that's one of the things that we're working on is trying to figure out what are the requirements out there, and then what are the right kind of technologies we can take advantage of to facilitate that?
HS: One of the systems that has been announced recently is a system called Jini, of Sun Microsystems, and that system (it's, by the way, spelled J-I-N-I for some reason, though it's pronounced "Genie") that thing imagines a completely distributed kind of computing environment with storage being anywhere you want and CPU's being anywhere you want, and that kind of thing. Is this the kind of thing that Internet2 is going to allow, to have my hard disk be in Michigan and my CPU be down in Tallahassee and --
JB: We've got a new concept of Pangaea coming along here, right?
HS: Well, I mean, that's the kind of thing that Jini sounds like it will do. And when I heard Ted talking about distributed storage, it sounded like I could have my hard disk wherever I wanted.
TH: Right.
HS: Not necessarily on my desktop -- I could have it in a sunnier climate!
TH: That's right. I think that actually Guy Alms, who heads the engineering side of the Internet2 project, says that profligate use of network capacity does not make an application inherently advanced. So --
JB: Too bad, Howard!
TH: So simply moving your disk into sunny climates probably doesn't count. But the thing is that the history of this mentioned earlier -- the history of NSFNet was tying universities to these very large supercomputers at universities like Cornell and locations like Pittsburgh and the San Diego supercomputer and the University of Illinois.
But the trend in the last couple of years has been by the supercomputer facilities both increased capacity on their sites, but also coming up with ways of linking together high-performance workstations around the networks to unite to solve certain problems. And this whole concept of computing is called metacomputing or the Grid or the Global Grid, which means that they think about the networks as being like the electrical utility system, tying together all these resources.
HS: So I just kind of sop up cycles.
TH: Exactly what they're trying to do. And there's some very interesting work being done by both the major supercomputing programs in this country who are part of the Partnership for Advanced Computational Infrastructure -- one program being led by NCSA and the University of Illinois, and the other one by the San Diego supercomputer center out at UCSD.
HS: Has anybody actually done that kind of thing, or is that still a thing very much in the planning stage, sopping up cycles?
TH: No, they are doing it, and they're getting better at it all the time. For example, there was a conference last week down in Orlando where they demonstrated people doing these kind of tasks across Europe, US and Asia -- of using (again, going back to the middleware theme) common security and scheduling software and so forth to be able to unite these resources together. They just had to sop them up and apply them to a problem. So, no, there's some very interesting work being done in these areas.
HS: You just mentioned security, and in Internet2, is there some additional security features or integrity features or things along those lines that make it a better network?
TH: Well, I would say that we're not really trying to do anything different as much as we're trying to make sure that we insure that security is part of the environment. That is, some people might say we might have to give up security in order to do real-time videoconferencing, but in fact, we've demonstrated that you can in real-time encrypt a video screen -- full-motion, full-screen video -- without interfering and putting in so much delay that it becomes unusable.
So I think that what we're really doing is finding a way to make sure that security features are available to application developers.
HS: I wonder if we could return to this Quality-of-Service issue. It's the thing that people talk to me the most about when they talk about Internet2. I mean, as soon as they get over the fact that, oh yes, it's a faster network and things, their concern is that for some applications, we really need Quality-of-Service guarantees. And I can understand how you might do it if there's just one connection that you're going to guarantee. But when you get lots of people wanting guaranteed Quality of Service, how are you going to do it?
TH: Well, that's a good question. We find that the technical challenges for determining which packets will get priority is daunting enough, but in fact, it will be the policy issues that will overwhelm that eventually --in that having to figure out who on campus does get priority? Is it the president's office? Is it the physics department? Or is it the music department with their need for high-fidelity audio?
And then there's issues of how do you make sure that everyone doesn't claim that their traffic is higher priority?
HS: Why wouldn't somebody?
TH: I think the thing that has to occur is there has to be policing mechanisms in the network that ensure that if there was an agreement between two universities or among a group of universities for a certain percentage of the traffic to be high priority, that the policing would ensure that you don't exceed those limits.
HS: Why not use economics as a way of policing the thing? One of the things that's been suggested to me many times was that if you just charge for priority. I mean, you want very high priority -- you pay a great deal of money. You want lower priority -- you pay less money. And that the economics of the thing would make it so that not everybody would want high priority because not everybody would want to spend all that money.
TH: And that very well may be the mechanism. In fact, using economic principles as a way of rationing and acquiring access to these resources very well may be the approach to doing this and is part of the whole investigation.
JB: Let me just mention at this point that it would be -- if any of our listeners have some questions they would like to direct to Ted on these important topics, that the e-mail address is expert@cren.net. Hope to hear from you.
Okay, going back to and leveraging off your idea of the economics, that is an important topic, Ted, and oftentimes it's a very difficult issue on campus to deal with. How do you in fact find the money to distribute programs that may in fact be very important, but there may not be the funding to do that, so certainly the sooner we can reach a solution on that probably will make policy-making a little bit easier.
TH: That's correct. The one thing is that it is a situation where the idea of a busy signal on the Internet is something we're not used to right now.
JB: Um-hum.
HS: Right, and that's going to be a real strange thing. You're saying that Internet2 might have busy signals, at least if you want something very high priority.
TH: That's correct. And what's the human behavior in that situation? People may just say, "Oh, okay, I'll just take a lower performing environment to be able to get my applications done."
JB: Well, too, like you say, if there's a balance between the cost -- if I can get medium priority for much less cost, I might redesign my application to be effective that way.
HS: Right. And I think there's obviously certain applications where you're going to say, "I can only have the best possible service or the thing's not going to work. I'm doing something that has very short response times and it's an expensive, delicate kind of thing. So I've got to get that, and I'll pay whatever I have to."
I assume, in talking about Internet2, that there must be some kind of Internet3 or Internet4 out there. Is there? Could you tell us about what follow-ons to Internet2 are being planned?
TH: Right. Definitely. The environment right now we're focusing on in Internet2 is a leading-edge but stable research network environment for doing applications development. And so you can imagine, there are certainly innovations occurring in the network space in terms of the physical hardware of the links that we want to exploit in the future. And so there's some work being done on technology such as dense Wave Division Multiplexing. It'll give us each higher capacity--
HS: Whatever that is.
TH: And I'll explain it!
JB: Good!
TH: We're talking about deploying in the next few weeks a network that's in the multiple gigabits-per-second, but with WDM, we can get into tens of gigabits-per-second.
The way we do that is right now, in most cases, you have a single laser sending a stream of information down a fiber optic cable. What Wave Division Multiplexing does is --
HS: Let's do this slow here.
TH: Yes. It essentially uses multiple lasers, each using different colors to carry information on each of those colors down that same single fiber optic cable.
HS: So you're kind of multiplexing with different colors?
TH: Exactly. And so the idea is you can increase the capacity of that fiber optic link without having to put in additional strands -- within a single strand of fiber optic cable. And so that will greatly increase the capacity of these network links. And start thinking about applications beyond things we're even talking about right now.
HS: The fiber just carries each color just as well? I don't know that much about fiber and colors.
TH: Right. It's true that we may not be able to do this across all the fiber in place. Some of the older fiber has interference at certain frequencies so that really it -- I'm learning more about this myself, but they've got this technology called non-zero dispersing glass that allows you to more efficiently use the full spectrum of colors. And that has been deployed sort of more recently. And so it really may be some of the new infrastructure that may be able to take best advantage of it.
HS: Is the Internet2 going to have any impact on the commodity Internet? It would seem like the folks who are on the commodity Internet are going to be screaming for some of the features and services that are available on Internet2.
TH: I think it definitely will. I think what will happen is that, as we learn about some of these things -- whether it's Quality of Service or multicast or using higher capacity links -- our vendor partners, the Cisco's and IBM's and Fores and 3Coms and so forth will build some of these innovations into their products and then make them available to everybody who accesses or builds commodity Internet connections. And that will raise the overall functionality for the broader Internet, which will then give us the ability to focus on that next set of Internet3 problems.
HS: So if we were to stand back, what you really imagine happening is that features from Internet3 are going to migrate to Internet2, and features from Internet2 will migrate to the commodity Internet. There'll be kind of like a pipeline that we have and things will just move from one to the other.
TH: Precisely.
JB: In terms of when you were talking about the fiber, Ted, and also then I'm thinking about the campuses and as they are getting ready for Internet2 -- so then one of the questions they'll want to take a look at is what is the state of the current fiber that they've got on campus and then be watching for the kinds of fiber that you're describing now that would allow for that Multiple -- let's see, what was it -- Wave dimension?
TH: Wave Division Multiplexing.
JB: Division Multiplexing. There we go.
HS: I'll use that in the next meeting I'm in.
JB: Anyway, so is there any advice you might provide for people planning campus networks as to how to approach that question about the fiber?
TH: Well, I think it is essentially that there are a lot of meetings going on within the Internet2 structure to talk about both campus and regional planning activities and learning from each other's experiences. And so if your university is an I2 member, you want to make sure that the information that is being exchanged in those fora are being distributed on your campus. And if you're not an Internet2 member, you may want to contact your regional network provider and see whether they've got connectivity to these kind of discussions.
JB: We do have some questions coming in, Ted, and maybe now is a good time to address them. There's one question coming in from Leslie Maltz from Stephens Institute of Technology, and she's asking a question about what application areas are not being developed currently right now that you think should be happening on I2? And if they're not being developed, what might be holding them back?
TH: Really, in my travels, I go around and visit different universities all the time. This week I was down at the University of Nebraska and I was down at Pitt and Penn State and so forth a week ago, and I've discovered there really are applications in every discipline that are under development.
I think that some of them may be underrepresented, though, and certainly the arts is one area I've tried to put emphasis on this year -- that is, trying to bring this kind of capability into the performance studio for dance or to masters' music classes and so on. But I think the thing that I'm most concerned about now is trying to move these from development into production. And we have an awful lot of applications being investigated and developed, and now we want to make them more widely available. I think one of the prime areas for that will be digital libraries.
HS: Talking about libraries, we have a message from Carla Garnham from the University of Wisconsin in Milwaukee, and she wants to know if you can talk about Internet2 applications that will have any impact on libraries and librarians?
TH: I would say definitely I've seen a number of things. For example, in using the digital video -- some of the digital library initiatives being funded by the National Science Foundation and NASA. There's a project at Carnegie Mellon University that's doing automatic indexing of video feeds. In addition, I've seen things such as the University of Michigan is scanning in its collection of papyri at very high resolution. And so these are large collections of images that if you wanted to go and sort through them and scan through them quickly, you want to be able to have them appear on the screen almost instantaneously. And so you want that kind of access. And the other thing is that you want to do this from university to university so we don't have to replicate each other's collections. So I'm seeing some very interesting uses from imagery to full data, or sorry, full text, moving beyond abstracts, that will be increasingly part of the Internet2 application set.
JB: Ted, going back to that example that you mentioned about the scanning in the papyri at University of Michigan. Are those files, then -- what size are those files likely to be, and is that why the Internet2 capability would be needed for those?
TH: I don't know precisely the size of the files off the top of my head, but they certainly can get out into the many hundreds of kilobytes and into the megabytes in size. And that's the kind of thing you could wait an awful long time to be downloaded over a dial-up link. And as a researcher, you want to be able to flip through them, not spend most of your time waiting.
HS: Ted, just going back to commodity Internet, I can't resist asking you the question that I think most people would have about the commodity Internet. We're talking about transferring technology from Internet2 to the commodity Internet. I think the thing that concerns most people about the commodity Internet is that it just takes so long sometimes to get Webpages and things to move around. Is there anything out there that's going to help that in the short run?
TH: Oh, I think on the commodity Internet there's some interesting work being done, for example, on compression -- to be able to take higher quality images and compress them into smaller files so they transfer more quickly. There's also things that people are doing in terms of trying to filter out the ads and so forth that fill up the pages. And I also think that there are some interesting technologies for home connectivity that are being increasingly explored, like cable modems and digital subscriber line technology from the phone companies that aren't full sort of broadband Internet2 capacity networks, but they're significantly faster than the current dial-up environment.
HS: Does that kind of thing mean that it might be useful at some point for people to use Internet2 from home? If you're a professor or you're a researcher or whatever, that you could be at home and you still use this thing?
TH: I think it's possible that we'll be able to do that. You won't be able to get the full-motion, full-screen video of MPEG2 quality.
HS: Even if you have a cable modem connection or ADSL connection?
TH: Well, because ADSL now is typically around a megabit-per-second and to do MPEG2 quality that we're doing now, which is about 4 Ohm megabits per second needed a faster link. And cable modems, definitely those network links are that capacity, but they're shared. So if everybody in your neighborhood's asleep and not using the network, then you might be able to do it, but you can't count on that.
HS: But let's suppose you're not going to do full-motion video, but what you're going to do is try to control this microscope somewhere. Do you have enough bandwidth to do that kind of thing from home?
TH: Not from home, not probably for at least five years.
HS: And in five years, what's going to happen that's going to change that?
TH: Things like the digital subscriber line will move up to six megabits-per-second from the one and a half that it is today.
HS: And that's what you think you need to be able to do some of this stuff?
TH: Yes.
HS: That you need that kind of bandwidth?
TH: Yes.
JB: That is all very exciting, though. When you said five years, you know, on the one hand that's a long time, Ted. On the other hand, five years goes by pretty quickly.
TH: Think about it, where we've come from and where we've yet to go. Next year, we'll have 2.4 gigabits per second networks in this country, and 11 years ago the state of the art in this country for national Internet backbone was 56 kilobits per second. And so there's a lot of change yet to come, still, too.
HS: So why should it take us five years to do this at home? To me, that seems like too long to wait. Judith, it might sound like a short time to you.
JB: Oh, I'm not very patient, but at the same time, looking back --
HS: To me, anything more than a couple days seems like it's much too long.
JB: Well, we'll go through three more generations of computers and chips by then, right?
TH: Most definitely.
JB: Yeah.
TH: The more competition we have among the phone companies, the cable companies, the wireless companies, the sooner some of this will be developed.
HS: I've noticed, Ted, that there's some stuff that's appearing on the Web now in 3-D. Is there any kind of Internet2 application that takes advantage of three dimensional stuff or stereoscopic kind of stuff, or is that just something that's part of some other application?
TH: No, absolutely. A lot of the applications that we're dealing with in the virtual reality space are stereo view. Whether you're looking at, for example, 3-D volume visualizations of the human brain from MRI data to whether you're looking at a model of a cyclone that was constructed from sensor data and from satellite imagery. So no, I think this whole space of what we call teleimmersion -- the ability to put yourself inside a synthetic world created in stereoscopic view around you -- is a major part of the applications interface in the future.
HS: Sounds great!
JB: That does sound great. We do have something else here that we, I think, want to talk about and that is you mentioned, Ted, at one point about Internet2 Days. That even for those campuses that are Internet2 members right now, that there's still lots of folks on those campuses that would like to know a little bit more about Internet2. Can you tell us about those?
TH: Sure. What we're trying to do is reach more and more of the faculty, and the thing is that faculty researchers and other kinds of people who come to the technical and engineering meetings, and so we have to go to the campuses and also to discipline workshops to try and reach them. So one of the things I've been doing is putting on these, we call Internet2 Days where we go to a campus -- [END OF SIDE A] -- own campus about what support services are made available to them. It's a way to really sort of energize the activity on a campus, and I've done this at, oh, I'm losing count now of how many universities across this country I've visited or have been put on by their own resources.
HS: How do we get you folks to come out and do this?
TH: You basically send me mail or give me a call and say that our university is interested in doing this, and I can guide you through what kind of resources are available to put it on through what speakers are available, to what kind of demos you might be able to put on at your campus.
HS: And this is free to universities? Because universities are always interested in things that are free.
JB: That's true!
TH: For the Internet2 member universities, yes, it is. And we also have worked with some of the regional organizations who then also invite some people who may not be I2 members, but they are part of that regional program and they may want to learn more about it too.
JB: That sounds like a good way to save both time and also leverage your time, Ted, doing regional kinds of things.
Well, listen, I see that our time has just really flown by for us today. And I'd just like to remind our listeners that if you've missed the early sections, that you can go ahead and the audio archive will be available within the next 24 hours. And also a transcript will be available by probably Tuesday next week.
Ted or Howard, do you have a final comment or question before we go into our closing notes?
HS: I see that we actually got at least one more e-mail question in. We're not going to have time to answer it on the air. I'd like to remind folks that we will try to get an answer on the Website. I know this gives you some responsibility here, Ted, but we will try to answer the questions that we were unable to answer. No, I just think that this has been a really interesting area, and I'm sure we're going to revisit this sometime in the future because it's moving pretty dynamically.
JB: All right, great. Thanks so much. And Ted, do you have anything that you'd like to mention?
TH: Just again to invite people to look at the Website to learn more about it and to feel free to send questions to me and we can follow up with more information afterward.
JB: All right, well, thanks so much. And I would like to thank all of our Web participants for being with us here today, and again, a reminder to go ahead and send follow-up questions to Ted at expert@cren.net. He'll be able to receive those for the next few days. And also I'd like to remind everyone that they can hop onto the net and hear Ted talk about Internet2 applications as part of the Virtual Seminar that's on the CREN site.
Be sure and mark your calendars for another TechTalk just two weeks from today, and that will be Mark Resmer talking about computers for everyone. Let's see. If you would like to receive announcement messages for the sessions, you can also send mail to cren@cren.net.
So thank you everyone today. Thanks to the board of CREN, to our guest expert Ted Hanss, our technology anchor Howard Strauss, and to Paul Bennett at UM Web Services for the encoding. And all of you for being here. You were here because it's time. 'Bye, Ted. 'Bye, Howard.
HS: Good night, Judith. Good night, Ted.
JB: 'Bye, everyone. Take care.
TH: 'Bye.
