TechTalks Transcript
Broadband on Campus: How to Get It and How to Pay for It
January 10, 2002
Audio
• Streaming
MP3
• Download
MP3 (Download
Tips)
 Judith Boettcher [JB] |
 Howard Strauss [HS] |
 Jeff Kuhns [JK] |
 Ron Hutchins [RH] |
JB: Welcome to the CREN Tech Talk series for spring of 2002 and to this session on Broadband on Campus: How to Get It and How to Pay for It. You are here because it�s time to discuss the core technologies for your future campus. This is Judith Boettcher, your CREN host and our session is coming to you today with the support of the CREN member institutions. If your institution is not a member, please join! Let us welcome the Tech Talk technology anchor, Howard Strauss of Princeton. Howard is a well-known web technology expert and portal expert and always comes armed with a fresh insight for us all.
HS: Thrills!
JB: Welcome, Howard. What kinds of fresh insights do you have for us today?
HS: I was actually going to announce myself as a radio station today because we�re doing broadband, but I thought better of it.
JB: Oh, what is it going to be, WPHS?
HS: I don�t know. If you look at Princeton University�s initials, maybe we don�t want to do this. Anyway, I�m Howard Strauss. I�m the technology anchor for the Tech Talk series of technology webcasts. In this webcast, I invite you to join Judith and me in a lively technical dialogue with our guest experts, Jeff Kuhns and Ron Hutchins, that will answer the questions you�d like answered about broadband access on campus and ask those very important follow-up questions. You can join in this dialogue by sending your questions via e-mail to expert@cren.net anytime during this webcast. If we don�t get to your questions during the webcast, we�ll provide an answer in the webcast archive. What is broadband anyway? Go ask some random people who work near you and you�ll get some surprising random answers. Ask the Federal Communications Commission and they�ll say�and I quote��broadband is a descriptive term for evolving digital technologies that provide consumers a single switched facility offering integrated access to voice, high speed data service, video demand service and interactive delivery services.� That�s perfectly clear, isn�t it? The broad part of broadband is pretty easy. It means the bands�whatever they are�are wide, not narrow. But what are the bands in broadband? It turns out that they are actually ranges or bands of frequencies measured in a unit called the Hertz, named after Heinrich Hertz who was the first person to demonstrate the production and detection of electromagnetic waves. One Hertz is one cycle per second, or one oscillation, often of an electromagnetic wave. Electromagnetic waves are what are used to transmit radio, TV, cell phone signals, the telephone and data traffic that is carried by microwaves and much, much more. I�d explain to you how electromagnetic waves make radio possible but Albert Einstein did it so much better. Here�s his explanation. Einstein said, and I quote, �Suppose there was a cat so big that his tail is in New York City and his mouth is in Los Angeles. Now, if someone pulls on his tail in New York, the people in Los Angeles will hear him meow. That�s exactly how radio works! Except there is no cat.� Radio, you know, has AM and FM bands. These bands are the same kind of bands that we refer to when we talk about broadband. You know that each of these bands is at some frequency. WQXR FM, for example, is at 96.3 on your FM radio in the New York City area. But 96.3 is actually 96.3 megaHertz, or millions of cycles per second, which is the frequency of the WQXR radio signal. But if a radio station broadcast only a single frequency, it would not make very interesting listening. Since a radio station carries information in the form of voice and music, it needs to add to the frequency of its broadcasting all the frequencies of the information plus it needs some extra frequencies to avoid interfering with adjacent stations. The radio station right next to 96.3 is 96.5, or 200 kilohertz away. FM stations are assigned a range of frequencies that are the frequency you tune plus or minus 100 kilohertz which results in a bandwidth of 200 kilohertz for FM stations. TV stations use a broader range of frequencies because they carry more information than radio. TV is broader band than radio. The width of the band in broadband is therefore a measure of the information-carrying capacity of a communication channel. If a university wants to move information around, it obviously wants the highest information capacity possible. But more bandwidth costs more money and costs are not dropping as fast as demand for bandwidth is growing. Like Einstein�s cat, the issues of getting and paying for broadband access on campus seem simple until we realize there is no cat! There are instead difficult hardware and software, regulatory and management issues on all campuses must face. In just a few moments, Judith and I will pull on the tail of the campus broadband critter and have Jeff and Ron�who are far away from us�see if they can make it purr on today�s webcast of Tech Talk. Judith?
JB: All right, thanks very much, Howard. That�s great to know how broad the bands really are. And we�ll find out from our experts today what they really think a broadband is and how we describe it. Let�s welcome our first expert for today, Jeff Kuhns, who is the Associate Senior Director in the Office of Telecommunications, which is a division of Computer and Information Systems at Penn State. Jeff�s responsibilities include budgeting, new business development and strategic planning. Jeff is completing his twenty-second year at Penn State and has seen perhaps those small bands grow into very wide bands. Welcome, Jeff.
JK: Thank you, Judith, and thank you, Howard.
JB: Good, and our second expert today is Ron Hutchins, the Director of Engineering and a research scientist within the Office of Information Technology at Georgia Tech. Ron is also a recent �doctor,� having received his Doctor of Philosophy in computer science just in August of last year. Ron�s current field of interest is computer networking in areas such as production network management, educational collaboration technology�always a hot topic!�high speed, large scale network design and management, certainly important to broadband, and also mobile and nomadic computing. Welcome to Tech Talks, Ron.
RH: Thanks a lot. I just wonder if Howard would tell us, is this a broadband cat or a narrow band cat?
HS: Don�t you understand? There is no cat!
JB: How large will the purring be?
HS: The trick to this whole thing is there�s no cat. Once we understand that, then we can understand the sound of one hand clapping and all kinds of other things here�
RH: Right.
JB: So no cat.
HS: What?
JB: So no cat?
HS: There�s no cat, right. That�s the trick here. Ron, when we say broadband, what do we really mean? Are we talking about Internet 2, are we talking about fiber, what is this thing, really? Can you kind of help pin this down?
RH: Well, I�m kind of like you described early on. I think broadband is a definition of bandwidth, not necessarily�I don�t think of it in hertz, but in, you know, for example the bitrate that you can send over the network would be a definition, I think, that�s kind of common. I�m also finding that, like you, everybody has a different definition of it. The telecommunications circles, folks that deal with telephones, 64 kilobits a second is, I think, a definition of narrow band and anything above that is broadband. Basically anything that�ll handle a voice call is narrow band, anything above that is broadband. That doesn�t mean a whole lot to me. My experience is that a 10,000 bit difference in bandwidth doesn�t make that much difference and my experience over the past 20 years is that it was not until we got to ten megabit Ethernet, away from the 9600 baud modems, that we noticed a real big change, so in a sense, an order of magnitude, two orders of magnitude difference is meaningful, so I�ll just throw out on the table that it�s more of a perception issue than it is a bitrate issue.
HS: Jeff, I�m always hearing about these OC 3�s and OC 4�s and all this other kind of stuff. Somehow this is related to broadband, right?
JK: Sure. The OC is just a two-letter definition for Optical Carrier, a way of defining bandwidth carried over optical fiber, so typically we�re looking at OC 3, although there are smaller levels. And an OC 3 would be the equivalent of 155 million bits per second.
HS: And how high do things go?
JK: Theoretically?
HS: What�s�okay, what
JK: Well, the Internet 2 has announced that they will�Abilene, rather, will be running at OC 192, so that will become a production service. I�m not sure what levels are currently running in labs. Ron may have a better handle on that.
HS: But when we take a typical university, okay, and say they�re not running Internet 2, what are they running? Are they running OC 3? When we talk about somebody running broadband at a university, what are they running?
JK: Well, it probably depends on how we�re defining that part of the network, Howard, and by that I mean that typically an institution�s going to have its own backbone network on campus. It may have a connection from the local community for students, faculty, staff, living off-campus, and then it may have a connection to the commodity Internet or to Internet 2 and each of those will probably be running at different amounts of bandwidth. So the backbone network on the campus may be an OC x network. Depending upon the size of the institution and the length of the connection to the Internet, that may be an OC x. The local community network is likely to be much smaller because the residential broadband definitions, as Ron was mentioning a minute ago, are much lower speeds.
JB: You were saying OC x a couple of times, Jeff. What about at Penn State right now? What�s the backbone on the campus and what�s the connection to the Internet?
JK: I�ll do it in reverse order, Judith. Our connection to the Internet right now is actually at an OC 3 level. We have two OC 3�s that go for commodity Internet and I2 connections. On campus, we have a backbone that�s an OC 48.
HS: And somehow you�re connecting to some ISP�s somewhere?
JK: Yes.
HS: We talked about some local loop connection.
JK: Right. The local loop is a term that�s sort of like broadband, has different definitions depending on how you apply it to a specific purpose. In this case, I�m referring to a local loop for Penn State being a line that connects State College, Pennsylvania�in the geographic center of the state�to Pittsburgh, some 105 or 110 air miles away. That local loop is what gets us to the Pittsburgh gigapop where we have both commodity Internet connections and Internet 2 connections.
HS: And what�s the speed of that local loop?
JK: In our case, there are two of those, each of those running at OC 3.
HS: Right, so it seems like it�s going to be�the slowest link is what�s going to be what determines what bandwidth you really get, so that local loop really has to be fast.
JK: Yes, because then there�s the commodity Internet that you�re�or commodity Internet traffic that you�re buying out of the gigapop which may actually be less than the local loop. So you�re right, Howard, you need to look at the smallest amount of bandwidth in your network connection as the ultimate choke point.
HS: And how do you decide to connect to Pittsburgh? Why do you connect to Pittsburgh as opposed to somewhere else?
JK: In our case, it was the most economical place to connect. And we�re very different than an institution like Georgia Tech that Ron can talk about. I don�t know that there are any two institutions in the country that are quite the same, but geography certainly continues to play a major factor in what you can afford to do and how you architect the network.
HS: Could you talk about that, Ron? What�s your broadband plan at Georgia Tech?
RH: Sure. Let me do one differentiation here.
HS: Sure.
RH: Jeff mentioned OC 48 backbone, OC 3 connections. There�s almost two competing camps and we actually call them the Bell Heads and the Net Heads. The Bell Heads are the folks out of the Bell operating companies who have been in telephony forever and then there�s the Net Heads. They�re the networking geeks on campuses that have only done campus networking. The Bell folks talk about the OC hierarchy, the SONET hierarchy. SONET is just the term that means Synchronous Optical Networking and the optical carrier stuff is part of that. The campus networking set, we talk about ten meg Ethernet, 100 meg Ethernet, gig E and ten gig E and stuff like that. And the two are not incompatible, they�re different ways of thinking about things. So let me start off by saying Georgia Tech has today a gigabit Ethernet backbone which doesn�t map to the hierarchy by it�s the data com side, same thing. No real difference if I talk about ten gig E or I talk about a C 192. I�m really talking about the same bitrate within some error bars. So does that make sense so far?
HS: Okay. And what�s that connected to? You were going to talk about your backbone.
RH: At Georgia Tech, like Jeff said, Atlanta is one of these cities that has rich connectivity based on a whole bunch of things, probably starting with the Democratic National Convention years and years ago when Bell South strung lots of fiber through Atlanta. So we�re one of what we call the NFL cities that has rich fiber communication and rich telecommunications connectivity. So my local loop is a mile and half of dark fiber that goes downtown to what�s known as a carrier hotel, which is a private building that�s owned by a company that has a whole bunch of telecommunications carriers in it, that once you get into there, you can connect up to anybody. Now, I�ll just make a generalization, and say for free. It�s not really for free, but it�s easier just to talk about. [inaudible] want to connect to and connect to them, any of them equally, so there�s a lot of competition and I can choose the one that gives me the best price. And my local loop cost is small, a small return cost. I have a lot of flexibility there, as opposed to Penn State which has just a few providers that get to Penn State. There�s not a lot of competition, there�s not a lot of effect on the pricing and it�s very expensive, so I can, for example, connect up to our Internet provider at�it�s confusing since Georgia Tech also runs the Southern Crossroads which is a gigapop in the southeast and the gigapop is in our machine room, so we connect up to that directly. But I�m going to talk about how the gigapop gets service. The gigapop is connected at OC 12 to our ISP and we�re connected up at OC 48 to Internet 2. And we service right now about 15 different members from schools in the [inaudible] region, the southeast, to that service, Georgia Tech being one of those schools.
HS: So what you�re saying is that capitalism or competition works to your advantage because there�s lots of competitors close by. And if you have the misfortune of being in a place that�s not close to one of those things, you�ll only have one or two vendors to choose from.
RH: That�s right.
HS: You�ll pay lots more.
RH: Yeah, geography is important here.
HS: Can we talk about the cost here? What does this thing cost? Or what�it sounds like you�re saying geography�s an important factor in the cost, but what are the other important factors here?
RH: Let�s differentiate a few things. First, I want to define what the Internet is. The Internet is, you know, Internet is like interstate highways. It�s highways that connect up states. The Internet is a network of networks, so there�s not really one Internet. It�s a whole bunch of networks and who do you want to connect to? You want to connect up to one of these players that has the most people that you need to get to, right? So what you do is you look around, you find the one that has the most people on it and it�s kind of like Beta vs. VHS. Once somebody has a whole lot of players on it, everybody wants to connect to them and so their prices go up. The players that don�t have a lot of people lower their prices to get more folks to come in and then they weigh those customers against the next vendor�s customers. The Internet is a very, very complex system of networks that interconnect at various rates. And they have private pairings and they have public pairings and that could take an entire hour to talk about that, all by itself. There�s also�the Internet itself touches down, these companies touch down at hundreds of places around the country, but not at thousands or ten thousands of them. It�s like I said earlier, the NFL cities are the places where all these guys touch down.
HS: Do you mean that literally, when you say NFL cities?
RH: That�s a term that�there�s some validity there. It�s basically the large metropolitan areas that�I would guess that 80% of them, maybe more, have NFL football teams. I didn�t make up that term. That�s just a term that the folks in industry talk about, kind of as a joke, but there�s some validity to the fact that if you�re large enough to be an NFL city, then you are probably a 99% chance a major metropolitan telecom area as well. So if we talk about the Internet, we talk about the touchdown points of the Internet, then all of us have to get to one of those touchdown points and that�s this local loop that Jeff was talking about. So you buy a service from somebody and then you have to connect up to it. So there�s two costs there. There�s the cost of the service and the cost of connecting to it. The cost of the service depends on the competition that you have to some large degree as well as the amount of service that you�re buying. You can buy a T1, one and a half megabits per second of Internet service, for some amount, but then you can buy an OC 12 for a whole lot less per megabit per second, although it�s a whole lot more money total. The amount you buy for incremental value is a whole lot lower, so it depends on the amount that you buy, it depends on the competition that you have. Then you have to add the local loop costs which also have competition in it. In a major metropolitan area, competition drives that price down, as opposed to having a single vendor.
HS: And it�s the local loop cost where geography has a real big effect.
RH: That�s right. Geography determines how close you are to these touchdown points. Geography also determines how much the loop costs you based on that and how many players you have to compete against each other to drive that cost down.
HS: Jeff, ultimately you�re going to connect to an ISP. Are there different kinds of ISP�s or are they all the same?
JK: Well, I�ll go back to something Ron mentioned a minute ago in talking about looking for ISP�s that connect to the most people in the most direct route, so clearly there are different kinds of ISP�s and in our case in Pittsburgh�which we use as an aggregation point�we can bring multiple universities together. We can then buy services from multiple ISP�s, trading off both the economics of that and the capability of having redundant services. So we�re trying to maximize the use of those dollars in that sense and yes, we�re looking at ISP�s differently depending upon the classification of the ISP as well as the service levels they provide.
JB: What happens
HS: Jeff��
JB: I�m sorry, go ahead.
HS: Okay, Jeff, a lot of small schools seem to be geographically disadvantaged here. I mean, if you�re a small school, then you�re probably not in a big metro area. So it�s going to be probably expensive for a small school. Why? Why does a small school need broadband?
JK: Well, I think there�s a couple of different issues I need to respond to there, Howard. First, I�m not sure that it�s true that smaller schools are necessarily in more rural areas. There may be a number of smaller schools in urban areas. There may be very large schools, like a Penn State, that happen to sit in a rural area. So it�s not always size��
HS: Good point.
JK: �that drives that. The reason why someone would want faster services, I think, is just driven by the increasing number of people who want to connect to places in a faster mode. We don�t try to track content at all and I hope we never do that, but certainly, we look at our students, our faculty, our staff increasingly having larger file size, starting from something as small as putting more and more attachments on e-mail to researchers who are collaborating across great distances to using multiple machines in a national or international network. So regardless of the size of the institution, I think all institutions are seeing the need for faster connectivity.
JB: Can you speak to some of the trends in terms of the growth and the demand for broadband or just simply high volume transfer rates?
JK: Sure, that�s certainly one, Judith. And another one that I think every institution has resident students recognize that every year you have a new crop of students that are showing up with the most current PC technology on the market. So you constantly have a generation of students arriving on campus with more powerful machines, students who have grown up increasingly using those machines for connections, perhaps from their homes, certainly from their schools, and it�s just second nature to them once they have a faster network connection, then, to look for places that have services they�re after that may have very large files. And that�s just looking at it from, if you will, the student as a consumer, not looking at it from the research side where there�s another major driver pushing the need for faster connectivity.
JB: When you�re doing�maybe Ron can answer this question, too�when you�re doing your planning out, you know, saying that this year and next year, what kind of growth are you predicting in your needs?
JK: We�re looking at growth that could be as high as doubling from year to year. In our internal campus backbone, we tend to blow out our networks about every three years and that trend�s been pretty consistent over the last decade. So we have the campus network factor that we try to plan for roughly three year cycles. Our Internet connectivity is the one that we can look at it doubling very easily year over year.
HS: Do you see the same thing, Ron, doubling every year at Georgia Tech?
RH: Yeah, we�ve been watching it closely since about 1995 and we have the records to show that it�s a very consistent doubling, about every 12 months. And the student body is staying the same, the size standpoint, so it�s in usage that we�re doubling.
HS: And what�s happening to costs? In a year when the demand doubles, what happens to costs?
RH: Well, it�s interesting. Costs are coming down on Internet service, but they�re coming down linearly and our growth is, of course, exponential. And even though
HS: [inaudible] over a long period of time.
RH: Right. It�s real bad! So we�re seeing problems with it. There�s a lot of things to talk about. We have found new funding sources on the campus for some of this stuff. The students are demanding Internet services and they come here expecting, you know, a technical university to have a high tech network and high tech access, so our dormitories are ten megabit Ethernet. I know a large number of the folks that we talk to at the universities have the same thing, and that�s per student. The students come in and expect that and when they bring in their machines, like Jeff said, they can push out 55 megabits a second with a heartbeat and they can�t get it, then there�s a problem. There�s another thing that�s an issue that we need to bring up, too, in why should universities keep growing their bandwidth. There�s a technical issue and the protocols that we use on the Internet�TCP specifically�that as you approach saturation on a network, as you approach the limit of your capabilities and you start losing packets now and then, those lost packets tend to cause sort of a tumble effect and your service just goes very drastically downhill as far as the quality. You don�t have to get far, you don�t have to hit your maximum rate before you have problems that are caused in your network. We�ve seen that. We tried at one point limiting our network bandwidth to a certain amount so that the students would back off but what happens is, when one student pushes it and hits that maximum, everybody�s bandwidth suffers and everybody�s quality suffers and we have a lot of complaints.
HS: I mean, do you see saturation at 60% or 80%? When do you see this effect push over the edge?
RH: I think it varies. I would say, you know, 80% for sure. I�ve seen networks that at 50% load were losing packets. It�s very, very hard to say. It�s very difficult. The network technologies today are not technically capable of moderating that, in my view. We�re getting there. We�re getting things like QOS is coming in and some of the vendors are talking about bandwidth limiting. We�re actually looking at bandwidth limiting at the edge of our network where everybody connects up rather than inside the network, and that will help us. But I don�t think it�s possible to say�I mean, there are people out there who know a lot more about this than I do, but I don�t think it�s possible to say, �Oh, you can hit 85% safely.� It�s going to depend on the type of your traffic, how bursty it is, on several different factors.
HS: Why is it that the prices aren�t dropping quicker? I mean, we can�t go on having demand double and the costs just go down a little and [inaudible] very far. But what�s preventing the costs from dropping quicker?
RH: Well, it�s probably several issues, but the first thing to ask is what is it that goes into the cost of your Internet service? And there�s several things. These Internet companies build the network, right? And the top tier of these Internet providers are spending hundreds of thousands of dollars on each of the routers and they may have 50 or 100 routers. It�s a lot of money. They also have to buy the connectivity that goes between these routers. That�s a lot of money. So the more customers they serve on this network, the better off they are, but then if they start serving high bitrate customers, then they start saturating their networks and as their networks get saturated, they have the same problems that we do with congestion and that�s a bad thing. So first thing is there�s a real cost associated with providing a network. Then, as I mentioned earlier, there are multiple of these network providers that interconnect. In the early days, these guys just said, �Hey, I�ll connect up to you and hand you traffic from me and you can hand me traffic from you and everybody will be happy.� And then it got to where some people were unfairly served that way because they were transiting traffic for everybody else and weren�t making any money off of it because they have no customers associated with that traffic. So they started charging what�s called piering fees. So these Internet service providers charge piering fees and they make money off of that. So the smaller guys have to pay piering fees to the larger guys for transiting traffic across their network. So that adds to the cost. So there�s a whole series of things that affect the cost. Some of the network providers today, for example, are saying, �We�ll charge you one rate for what�s called on-net traffic for anybody that we deliver traffic to for this on our network and another rate for other people.� And you know, on our campus, we don�t have any charges associated with delivering traffic across our campus right now so there�s multiple different prices. Now, the network pricing is going down linearly for a number of reasons. Telecom services are going down because there�s a lot of competition out there. The price for equipment is coming down in relationship to the bitrate that it�ll handle because of the wonderful things about Moore�s law and because competition just across the board is heating up on these Internet providers. So that�s the good news. The bad news is that our demand doesn�t seem to be leveling off. We keep finding new killer apps with higher and higher bitrate needs, you know, audio, music takes more than text. Video takes more than audio and who knows where it�s going to go from there?
HS: Ron, there�s a broadband pricing working group. Sounds like they�re trying to do something about this. Could you tell us about them and tell us what they�re doing?
RH: Sure. And Jeff should feel free to add in here, too. Jeff and I have both been involved in this.
HS: Are you both members?
RH: Yes.
HS: Okay, so we have two members of the broadband pricing working group.
JB: And maybe as Ron and Jeff are talking about the working group, those of you with questions out there, it�s a good time to send them in to expert@cren.net. Okay, Ron, go ahead.
RH: Well, Jeff, maybe you want to start us off since Penn State was really instrumental in getting the Broadband Pricing Group started.
HS: Sure.
JK: Be happy to, Ron. Actually, we were trying to do something, Howard, that fits very well with what Ron was just talking about. A number of us got together a couple of years ago to try and talk directly to carriers and ISP�s about the pricing models they used. We felt that, especially on the carrier side, they were still using pricing models that were telephony based so regardless of how wide the pipe was, many of them would divide it into the number of 64 kilobit channels they could derive and the pricing models were based upon those things. So we started out trying to convince them there might be better models. We�ve evolved a bit into now what we�re trying to encourage them to do is what Ron would call disruptive pricing models and we�re still working in that sense. We�re also trying to work as a group��
HS: I just can�t [inaudible] this. What�s a disruptive pricing model?
JK: Disruptive to the local models being used currently, so anyone who would come out and say, �We have a better way to price this. We think, for example, that our network should run gigabit Ethernet and we will price in chunks of 1,000 megabits or 10,000 megabits.� A different way to price, perhaps using different economies, and giving us a better price in the long run.
JB: So the institutions are proposing back to the carriers this type of model, is that correct?
JK: We haven�t actually gone with any model. We talked about that in the early days, never quite did that. But I think what we discovered, maybe just through serendipity, is that by holding these conversations, we found a number of new market entrants who were willing to listen and who were willing to come to us and say, �We think we have a better way to do this, from a pricing standpoint.� That�s been very, very beneficial. We�ve also learned that, not back to your cat analogy, but there are lots of ways to skin this particular cat. And so we have a number of institutions who have taken different approaches to provide what they consider affordable broadband and we have a number of those listed on our website, so I would encourage people to go out there and look and read through some of those examples. They may not fit exactly with another institution or another location, but it may give them ideas of what they could do in their own local are.
HS: Okay, what�s the future of this group? I mean, where are they going? Do they have some new initiatives or��
JK: Well, we�re headed to Phoenix in about three weeks, but I suspect that�s not quite what you meant.
HS: [inaudible]�
JB: They have broadband out in Phoenix, right?
JK: We continue to meet, typically three, four times a year and Ron, I�ll use that as a segue to pass the baton to you.
RH: Okay. There�s two issues, I think. One is exactly what Jeff said, which is new models for pricing that can help us to lower pricing. The other is, I see education as a market in and of itself that very seldom holds together to get benefit from that, maybe. One of the things we�ve been trying to say is, �Look, education has value to industry and we want to teach you that value so that you see that by helping us, by lowering your prices to us, you get benefit.� And we�ve actually had some successes there. I think where we�re headed in the future is to continue the two things, to continue to look for ways to help industry find disruptive models or new models that will help lower prices, and number two, continue to work as a community to let our value be known, to attract vendors to the table to us with lower competitive prices.
JB: You��
HS: Ron, how does all this wireless stuff that people are using affect the broadband picture?
RH: Well, again, I think broadband as a definition doesn�t depend on whether it�s over copper or fiber or wireless or barbed wire. It depends on the bandwidth. The wireless applications in general are the same applications that have just been made mobile or, you know, able to move around or whatever. The difference to me is wireless uses the same frequency spectrum as what�s on a fiber but there�s only one of the wireless frequency spectrums within basically the line of sight where you�re working. So people can interfere with each other and you have to be careful about that. There�s also security implications there that are very large with wireless. Other than that, I don�t think it�s really important to differentiate between the two.
HS: It seems that wireless, though, is narrower band than the things that are not wireless. Is that true?
RH: Well, not really. Like I said, the spectrum is the same. Now, the FCC manages the radio spectrum across the country and they cut it up in chunks and they deliver chunks, like you said early on. FM radio has a chunk that�s�you have a lot, I think you said, 200 kilohertz.
HS: Yeah.
RH: And TV is a wider channel than that. The wireless radio systems have�they started out�there�s lots of different wireless radio systems. The ones that have been the most popular have been the ones around 900 megaHertz, 2.4 gigaHertz and 5.6 gigahertz. As you go higher up in the spectrum range, they give you a wider chunk, so the 900 megaHertz stuff was around ten megabits a second. It was like an Ethernet. The 2.4 gigaHertz stuff was about the same. But the new stuff, which is standardized under the IEEE 802.11A standard is, I believe, up in the 56 megabits a second range. I�m not terribly up on all the small details but it�s much higher bandwidth and does fit my definition of broadband very clearly.
JK: There�s also some other spectrum that�s available and acronyms you might here like LMDS�Local Multipoint Distribution Systems�that will provide hundreds of megabits per second and people that are interested in that might want to look at Virginia Tech�s site. They actually have a couple of LMDS licenses and have been doing some interesting work. Typically, these are services that are point-to-point, run a few miles, but they might be a way for people to connect distant sites back to their campus or, if they�re lucky enough to have a line of sight to a gigapop and they can do it wirelessly, then they could consider that.
JB: Would they be using something like that for videoconferencing, Jeff?
JK: They could, Judith.
JB: Okay.
JK: I don�t know all the applications that Virginia Tech is running, but the bandwidth would certainly support it.
HS: Do we have any sense, Jeff, of how widely broadband is available at colleges and universities? I mean, do most colleges and universities have broadband?
JK: That�s my sense, Howard, but that�s not based upon any formal study. Just from talks and reading. I think most universities now in terms of their resident students follow the pattern that Ron mentioned where the dormitories have been wired for Ethernet to the bed or what some people call Ethernet to the pillow. Most universities are connected with fairly high speed links. They may not be an OC type link, they may be 45 megabits DS 3 type link. But I think increasingly, schools have that level of connectivity.
HS: Well, if most schools have that already, then what�s the broadband challenge for these schools?
JK: Maintaining, as we mentioned earlier�meeting, rather, the ever increasing demand for more and more bits per second.
JB: What about in terms of the costs on that? I think we�ve mentioned a couple times about the demand doubling but the costs only going down linearly. What advice or how are campuses meeting this challenge?
HS: Yeah, how do we pay for that?
JB: Yeah, how do we pay for it?
JK: That�s the $64,000 question!
RH: One opportunity is we can start selling the land on the land grant universities to pay for it, but that�s probably not a scalable solution.
JB: [inaudible]�
HS: [inaudible] all the land, right?
RH: Right.
HS: The virtual university, then.
RH: Yeah. You know, pricing just ranges across the board. Like Jeff said earlier, there are a lot of different kinds of ISP�s. We call them Tier One, Tier Two, Tier Three. I don�t know if those tiers are real meaningful nowadays. They used to be when the Internet was first formed, around the NSF Net commercializing. But it�s basically you go by the percentage of the total number of Internet routes that are carried by this network and you know, one provider, for example, holds 70%, one holds 26%, one holds 3%, one holds 1% and so forth. And they charge commensurately to their visibility, I guess you�d say, so the prices range anywhere from�if I take it on a one megabit per second, paid for a month at a time, so megabit per second per month is the unit that I want to be using, it�s like furlongs per fortnight, if you remember the math in college.
JB: One megabit per second per month?
RH: Right, one megabit per second per month. So if you have a ten megabit Ethernet to your provider, that�s ten megabits per second, paid for on a monthly basis, would be ten megabits per second per month. So if you divide it down to try to do apples to apples and we�re talking about one megabit per second per month, the prices range anywhere from $700 a megabit per second per month down to, I�ve seen as low as ten dollars
JB: Ten dollars?
RH: Yeah, $10.00 per megabit per second per month. Now, you only get those prices, the best prices at the highest quantities. Maybe you have to buy a gigabit per second total bandwidth in order to get those prices.
HS: So the bigger schools get the best prices, or the schools that are using this the most?
RH: Right. And also, that best price means that you�re paying for the local loop to somewhere and the company that�s providing it doesn�t have to pay anything to get you to their router. So it�s hard to do apples to apples comparisons here.
HS: But that sounds like big schools have a real advantage, then?
RH: Well, if we didn�t have so much demand and it wasn�t costing us so much, I would say yes, but
HS: But you�re saying the unit cost for large schools is smaller?
RH: The unit cost for high quantities is smaller. Now, what�s going on now, you�ve heard Jeff talk about the Pittsburgh gigapop, you�ve heard me talk about Southern Crossroads. A lot of our universities are partnering together in regions across the country and jointly buying services, so they�re leveraging the aggregated quantities between all the universities to get better pricing.
HS: So that�s what you�d recommend for small schools to do, is to go out and form coalitions with larger schools?
RH: I think there�s
HS: Well, maybe with other small schools, but just to get together.
RH: Yeah.
JK: There may be lots of ways to approach that. In some states, there may be a state correlation that already exists that they could enter into, or they may need to create one of their own. The one thing I think we�ve learned from our meetings as the Broadband Pricing Group is that there simply isn�t one approach that is acceptable everywhere or that works everywhere.
RH: Right. That�s a key to take away is every situation is slightly different. But understanding the basic concepts, that the higher bitrate you buy, the lower the price, the easier time you have of getting to high competition areas, the lower the price. Those are the key concepts that you can take away to try to figure out on your own what your best solution is.
HS: You say that most colleges and universities have broadband access, but that broadband access is not equal, right? Some universities have broader band access.
RH: Right. If you look at it on a�I�ve never done this, but just thinking, if I took the amount of bandwidth that we get into Georgia Tech and divide it by the number of students, that gives me some number, and that number may be high or it might be low. And that may be one way of looking at it. Some universities are going to have lower quantity per student if you divide it out like that. It may be a meaningless term. Like I said, I haven�t thought about it a whole lot, but just give me the benefit of the doubt for a second. That�s one way of looking at it. The other is, once you get to that point, the quantity of traffic that each student uses may be different based on their area of study. Some areas of study are going to demand a whole lot more bitrate than other areas so some schools, people send e-mail. At other schools, people will videoconference and the two are very, very different in the bitrate demand. There�s probably a third issue, which is how close you are to the peak of your bandwidth needs and the quality of the service that you can get based on that. At some schools, they�re in the position where they�re at the top of their bandwidth consumption platform and the quality is degrading because of that.
HS: When do you decide to use Internet 2? I mean, do you make an Internet 2 decision based upon we need more and more bandwidth or because we have some special application or how does that fit into this picture?
RH: Well, I think it�s two different things. As I mentioned, the Internet is a whole bunch of different networks and they�re connected together. Internet 2 is a research network that does not peer with the Internet so it�s a separate thing. And what you look at, as we mentioned earlier, is what�s the content on there or what�s the connectivity on Internet 2 that I need? Do I need that? Do I need high speed connectivity to another university or an Internet 2 partner to do research or to swap any kind of traffic or is there other content on Internet 2 that I need to get to? That would be the differentiating factor. The good news from my standpoint is Internet 2 as a network has been built very robustly so that the quality of service on Internet 2 is just tremendous compared with the Internet today. There�s a smaller number of people on there, there�s a smaller number of connections, it�s a very high bitrate network and the quality�s great. Now, I�m not going to get to Microsoft to download my patches for NT over this network because they�re not connected. But if I need stuff that�s on Abilene, then I can get just phenomenal quality in doing that.
HS: But you said�one of the things you said is, you said, �Gee, it�s wonderful! There�s not too many people on it!��
RH: Who is it that said, you know, that restaurant�s no good, nobody goes there anymore because it�s too crowded?
HS: Yeah, I think that was Yogi Berra, actually. And if he didn�t, he should have!
JB: Right. We have a question��
HS: We�re going to have the same sort of phenomenon with Internet 2 where if a lot of people go there because it�s not crowded��
JB: We�ll soon be at that, between 50 and 80% that we won�t know what will happen, right?
RH: Right.
JB: It looks like we have a question coming in related to the geography and the idea of really aggregating schools together jointly by services. It�s�the question is coming from Peter Marshall from Canada, from the [inaudible] group. So he�s asking, �Why not aggregate all schools [inaudible] by services.? Could you do it on a country-wide basis, all your schools���
HS: He�s thinking of Canada, of course.
JB: Of course! Well, �and then use networks like Abilene as the transit to the cheapest place in the country?� Any thoughts from either of you or both?
JK: Well, I think my thoughts, Judith, would be to say I think it�s a great concept. Sure, we should be able to do that. Practically, I�d be very surprised if we were able to pull that one off. Too many different situations, private institutions, public institutions, state-owned institutions, state-related institutions, people who want to collaborate, people who don�t. I just don�t see practically how you�d ever make that work.
JB: But we do have some examples, perhaps, on a statewide basis, don�t we?
JK: Well, we certainly have��
JB: --that might be comparable to that?
JK: Sure, and UCAID with its Internet 2 project and its Abilene network, I think, are examples of that sort of concept. But trying to get every higher education institution in the country to agree to be part of something and then be able to politically put out, for example, an RFP and select one or two winners sounds like an arduous task.
JB: I see. What about other advice in terms of or strategies for paying for broadband if we�re going to continue really growing these killer apps? And once people find that they can work really fast on the networks, they love it and they want to go back and do more. What thoughts do you have as far as for institutions? How do they pay for it? Can they get it paid in a consistent fashion or do they have to wait for special funding? Ideas or thoughts there?
RH: I attended a conference out in California, one of the SCENIC conferences one year. There was a large discussion. I led a panel discussion on pricing and how to pay for network services, and again, just like there are many kinds of different networks, there are many different ways of paying for this stuff. Some folks bill back like they do for telephone services on a per-port basis. Some folks have an off-the-top allocation to pay for the network. In some cases, the students pay. To me, the more important issue is, if it costs you a whole lot to bill, then you�re raising the price and the right answer is to find the balance between the administrative overhead of billing and the value of collecting dollars. I don�t know if there is one right answer. There�s probably a whole bunch of right answers, but it�s a complex question.
HS: What�s happening in the whole broadband area in terms of increasing capacity? You�re saying your demand doubles every year and your [inaudible] ISP�s are on some network. It seems that their capacity has to be doubling every year, too, then?
RH: That�s right, and again, there�s two piece of this�or maybe more. There�s the ISP�s themselves have a backbone that has to grow and then the ISP�s that are connect with pierings that have to grow. And most of these guys, when you buy service from them, one of the key things is to get them to disclose to you what their pierings are, who do they pier with at what rate. That�s one of the key elements, that�s the things that are most closely held by the ISP�s. They will not disclose them except under non-disclosure and then only under a lot of pressure because that�s their life�s blood is who they connect with and what speeds. And they want to grow those things only when they have to because it�s high cost. And you want them to grow them before they have to because that�s the quality of your network. Those are the two major elements of cost. Upgrading the machines, like Jeff said on his campus, every three year cycle, they upgrade the backbone. That requires an equipment change out a lot of times, a forklift upgrade and the forklift upgrades are pretty expensive.
HS: But is there any technology like wave division multiplexing or something like that that�s coming that�s going to help relieve the problem?
RH: I think wave division multiplexing is going to help for probably a couple reasons, and again, there�s no clear answer here. What are the high cost things? The high cost today is digging in the ground and putting in fiber. There�s a lot of fiber in the ground and they�re amortizing the cost of that fiber. Wave division multiplexing is like taking the white light going through a fiber and making it into a rainbow where each color can be used separately and independently and what that means is that that one fiber can be used to generate, you know, 128, 256, 400 separate circuits that are [inaudible]�
HS: So we take the same fiber and suddenly it can handle 400 times the capacity?
RH: That�s right. But it also means that you have to change out a lot of the points. Again, distance is your enemy here. Every 200 kilometers or so, you have to put equipment in that fiber that regenerates the signal and the equipment to do wave division multiplexing is more expensive. I�m generalizing a lot of things here. It�s not always 200 kilometers. There�s a whole bunch of interim steps that have to happen that would take us a while to get into. I�m not an expert there anyway. But just take it to say that wave division multiplexing will lower costs in some ways by increasing the capability, but it adds costs in other ways by making us change out equipment.
HS: But are there other technologies out there like WDM or other things coming down the pike that are going to help here?
RH: Well, my favorite one is tangled photons. Have you heard this one?
HS: No. But I guess we�re going to!
RH: In some of the large federal laboratories
HS: You heard it here first!
RH: --they�re experimenting with taking a beam and splitting the beam, turning the rate down until it�s a single photon gets split, then send it off in two directions and what you do on one of the photons exactly, something happens on the other photon at a distance so it�s action at a distance. And I�m waiting for that one to get commercialized.
JK: I think the caution, though, Howard, would be to simply have a strategy that says I�ll wait for technology to bring down the cost is probably not a good strategy.
RH: Right.
JB: Looking forward, what�s on the agenda for the Broadband Pricing Group work coming up?
JK: Go ahead, Ron.
RH: I was going to say I don�t know exactly what the agenda is yet, but Jeff, you may.
JK: One of the things we�ve been trying to do that may sound odd, but we think it has an implication here for pricing, is to get a little bit more involved in the policy arena. And in order to do that, one of the tasks that we�re trying to work on is to make a long range projection�long range perhaps of five years�of where we think we should be positioned at that point in time. If we can see a goal that we�re headed towards in terms of bandwidth, in terms of technology development, that should enable us to set some directions for policy that we could feed back then as recommendations or goals to Educause.
JB: You would actually take a look and see what the demand and the applications driving the demand might be five years out?
JK: That�s part of it, sure. Where do we think we need to be in five years from a networking perspective and if we have a fairly clear picture of that, that should be able to help us work with policy makers then or at least come in and have a position on policies that are being developed. Whether it�s something as simple sounding as should there be competition or not to other policies that are being kicked around now about tax relief for broadband providers. It�s difficult to deal with those one at a time in a short range viewpoint.
JB: It apparently sounds like those kind of issues take a long time to work anyway.
JK: Yes.
HS: Robert Singles from University of Minnesota just asked what the web address is for the Broadband Pricing Group, but do we have that on the website, Judith?
JB: Yep, it�s right up on the events page.
HS: Robert and anyone else who�s wondering where they can find more information about this, it is on the CREN website for this talk here. Jeff��
JB: Did you want to ask a second question there?
HS: Oh, I�m sorry! I didn�t even see the second question there!
JB: They always have two questions!
HS: I�m sorry, no one ever sends us a single question and if they do, we just won�t accept it! But yeah, Robert�s second question is, �Does the group have any plans for consortia to negotiate prices?��
JK: Well, the group itself, the Broadband Pricing Group itself doesn�t, at least as far as I know, but there are a number of consortiums that are members of the group, and Ron, you�ve mentioned soon that you might want to talk about that a bit.
RH: The Broadband Pricing Group has been walking sort of arm-in-arm with another group called the Quilt.
HS: The what?
RH: The Quilt.
HS: Q-U-I-L-T?
RH: When we say the Quilt, we have this vision of several ragtag pieces of fabric loosely stitched together, right? Well, that�s probably a good description here. There are�one of our rules is that hierarchy is your friend. We have campus networks, we have state networks, we have regional networks. The regional networks basically have been revamped since the Internet 2 came along to aggregate traffic to go into Internet 2 and those regional networks have been meeting together for, I guess, three or four years now and have put together an association which is a project under Internet 2 just like Abilene is, called the Quilt. And the Quilt has been in a process to try to, as a consortium, get pricing down for ISP services to the Quilt members who then pass that directly on to their members who are interconnected and that�s one of the things I mentioned earlier. Pittsburgh and Southern Crossroads both are selling ISP services to their members.
JB: Do we have a Quilt website listed, linked off of anyplace right now, Ron?
RH: There�s a Quilt website, www.thequilt.net.
JB: Dot-net, okay.
RH: Yeah, quilt.net is for quilters, people who actually make quilts.
JB: They would probably take umbrage at the tatteredness that was in your description, Ron, but we won�t go there.
RH: Well, my grandmother used to make quilts. I still have one of her old ones and I�m very fond of it.
JB: There you go!
HS: Am I in the wrong session here right now?
JB: We went off in the wrong direction, that�s right.
HS: If we have a university and they�re trying to figure out if they have an adequate broadband access, how do they know? I mean, how do they know when things are okay and when things are not?
RH: They probably know very quickly when things are not.
HS: Because? Students
RH: Yeah, because, well, students and, I suspect, faculty and staff members, they�re probably getting quite a few people showing up at their door or sending them nasty notes or calling them with nasty comments.
HS: Telling them things like, �The Internet is slow�? I mean, what kind of things are you going to hear when the broadband access is bad?
RH: Can�t repeat some of them in public, but
HS: But what�s the sense of what you�re going to hear?
RH: Yeah, the students�it�s interesting that over the Christmas holidays, no faculty, staff or students were on campus, basically. Well, I guess we had some students on campus but no classes were going on. We had students who would call up our Operations Division at three a.m. saying, �The network is down,� or �The network is slow.� So students are on all the time, always doing something. They care about what they�re doing, it�s important to them, and they let you know when a site is not reachable, when the bandwidth is slow, when the number of errors on their machine are high, all those kinds of things. They�ll let you know.
JB: Okay.
HS: So as long as you�re not getting those calls, you sort of think you�re okay?
JB: Well, as long as you�re not taking the calls by the network and the network is down, then�that would be a bad thing.
JB: Okay, well, we�re very close to the end here and we will have to bring us to a close. Do either of you have a final comment or a final message or thought that you�d like to leave with our listeners today? Jeff?
JK: Nothing of great note, Judith, other than to say that this really is an evolving process. It�s a process that differs from institution to institution, but I would encourage people that have done innovative work or think they have found a great solution or have found a carrier or an ISP that is willing to work with them to let the rest of us know and the easiest way to do that would probably be through the Broadband Pricing Group.
JB: Well, that�s right, and the website for the Broadband Pricing Group is nice and has lots of links and stories there, so that would be a way to share their knowledge then, right?
JK: Correct.
JB: Okay, Ron, what about a wrap-up thought from you?
RH: I would just say that in ISP pricing, you earn every dollar that you save by the work that you do to pursue that. It doesn�t just happen to you.
JB: Okay, sounds like there�s some real��
HS: Really go after those people.
RH: Yep.
HS: It�s good to know that you can negotiate with them.
RH: This has been great fun for me, so thank you folks for hosting us.
JB: Okay. Howard, what about your final question?
HS: Actually, I see it�s one minute of and any question I have is going to take a lot of time to answer. I think that at the onset we said we were going to try to get this broadband critter to purr and I think we�ve gotten pretty close to that. I�m really pleased, too.
JB: Okay, very good. With that, we invite everyone to explore those links on the website a little more thoroughly and you can, in fact, send follow-up questions that you perhaps didn�t have time to formulate to expert@cren.net and our experts occasionally will be able to answer them following the session. So thanks so much for being with us here today in our first session of 2002. Join us again in two weeks, on January 24th for a session on How to Organize and Respond to Security Events on Campus and our guest expert will be Kathy Kimball, the Security Officer at Penn State. So many thanks to our CREN member institutions for support of today�s program; and many thanks to our Tech Talk experts today, Ron Hutchins and Jeff Kuhns; to Technology Anchor Howard Strauss; to Terry Calhoun, Tech Talk web guru who persisted in the face of great bugs; Jason Russell, Gayle Terkeurst and the support team at Merit Network; to Susie Berneis, the audio file transcriber; and finally, a thanks to all of you for being here.
END OF WEBCAST
