How the Internet began in New York State, the current state of Internet2, and the remote possibility of Internet3
Ben Chi was involved in the first implementation of the Internet in New York State. He is Director, Advanced Application & Network Technology, at NYSERNet (the New York State Education and Research Network) and also Director of Computing Center Emeritus at the University of Albany.
UBIQUITY: Let's start by talking about NYSERNet. What is it? Why was it formed?
CHI: NYSERNet, the New York State Education and Research Network, is a not-for-profit corporation that was formed in 1986 to provide access to the Cornell Supercomputing Center by research institutions such as NYU, Columbia University, University at Buffalo, University of Rochester and several others. It was, essentially, the first manifestation of the Internet in the State of New York. It operated at 56 kilobits per second, a very high speed at that time.
UBIQUITY: How difficult was it to get the network up and running?
CHI: There were some problems at first. NYSERNet brought in partners from the commercial world such as New York Telephone to get the network running, NYTEL initially having an operational role. A year or two into it the people who first developed the network found New York Telephone to be temperamentally ill-suited to operation of an experimental network and so NYSERNet itself took over its operation.
UBIQUITY: What was the nature of that temperamental inability?
CHI: The people in the telephone world at that time had an obsession for ultra high reliability, this to be provided by hard-wired circuits. They had little interest in implementing new protocols or other concepts required by a radically different network design. Telephone networks are implemented with wires and connections from originating point to destination point. Whereas, IT networking is packet-oriented rather than connection-oriented, making use of multiple and varying paths. You throw a packet into the network "cloud" and somehow it gets to the destination point via one or another path. Successive packets might take different paths. They might even arrive in the incorrect order so there has to be ways to keep track of arriving packets. It's a different way of thinking about communicating data from point A to point B.
UBIQUITY: What happened after NYSERNet took over responsibility for managing the network?
CHI: A year or two after NYSERNet took over it became apparent to some people that there might be a commercial opportunity in its technology. They proposed to NYSERNet's Board of Directors that they buy the network assets from the corporation and operate it as a for-profit corporation. This commercial spin-off from NYSERNet was called PSINet, formed in 1989.
UBIQUITY: What does PSI stand for?
CHI: PSI is an acronym for Performance Systems International. Bill Schrader and Marty Schoffstall were the two who formed PSINet, though Schoffstall left the company in 1996 to form a venture capital firm. PSI became a very large and successful company that, however, took a nosedive as did so many others in the dotcom collapse and for the same reason as so many others -- by acquiring excessive debt. NYSERNet, meanwhile, continued on with its purpose, that being to facilitate and promote very high performance networking in support of the research and education mission of higher institutions in New York State.
UBIQUITY: What is the current configuration of the network?
CHI: We operate an OC-12 backbone across the State of New York along the New York State Thruway with PoPs, or connection points, in Buffalo, Rochester, Syracuse, Albany and New York City. This distributed gigaPoP architecture is the New York State component of Internet2, the nationwide research network.
UBIQUITY: What are some of NYSERNet's other projects?
CHI: We started a second project in response to high local loop costs. The local loop connects a participating institution's network to the backbone and is usually provided by the telephone company. It's expensive and becomes even more expensive the faster you need it to go. In Manhattan, in particular, where we have a fairly high concentration of participating institutions, we were tasked by our Board of Directors to explore the possibility of developing a replacement for the local loop. We found that it was not only feasible but also potentially attractive financially to install our own optical fiber under the streets of Manhattan to take the place of the local loop. While doing so, we could also implement private networks for institutions that had multiple presences in New York City such as the City University of New York, which has many campuses over the island. Operating one's own optical fiber has the additional advantage that bandwidth upgrades require only a one-time expense for new terminating electronics rather than increased monthly usage charges.
UBIQUITY: How was the plan implemented?
CHI: The plan has two parts to it: an optical fiber loop circumnavigating all the sites of a participating institution, it being a loop for recovery purposes, so if there's a breakage at a certain point you can go the other direction. A second loop connects the primary site of each institution with NYSERNet's co-location site, which is a central hub in downtown Manhattan where the institutions can access both the research network and the regular commodity Internet.
UBIQUITY: What do you consider peer institutions of NYSERNet?
CHI: NYSERNet is just about the only regional network that has survived from the beginning. There are regional entities like SURANet, the Southern Universities Research Association, which is a consortium of research institutions. Merit in the upper Midwest has more of a commodity Internet slant than NYSERNet, whose core business is a research network. In the northeast, you have NOX, which brings a number of New England institutions to a hub in Boston that connects to Internet2. A totally different entity is CENIC, the Corporation for Education Network Initiatives in California, which provides high-speed networking to the U Cal and Cal State campuses, the major private universities and the K-12 school systems.
UBIQUITY: What is the current state of Internet2?
CHI: Internet2 is at this point a mature, nationwide institution. Its mission is to support research and educational activities. Typically, people use it for very high speed file transfers and applications that require real time responses -- the sort of things you can do over an Internet only if the reliability is rather high and the bandwidth is very high.
UBIQUITY: What is the bandwidth of Internet2?
CHI: The New York State backbone is OC-12, which is 622 megabits per second. That is being replaced by the faster facilities that are able to signal at 1 or even 10 gigabits per second. The current production bandwidth tops out at approximately 10Gb/s. The national network to which NYSERNet is connected is being upgraded to 10 Gb/s.
UBIQUITY: Internet1, as we can call it now, became commoditized. Is the same thing expected to happen with Internet2?
CHI: I don't think so because there isn't a need for commoditization. Previously, there wasn't an Internet zero to supersede. The whole emphasis and thrust of Internet2 is very high reliability and prompt throughput, which are not needed by applications like Web browsing and e-mail but are needed for applications such as telesurgery and teleconferencing, which operate on a real-time basis.
UBIQUITY: Are there any strong commercial uses for Internet2?
CHI: No, and that's pretty much by design. The conditions of use of Internet2 don't allow for those kinds of applications. That's not to say that commercial entities can't participate but they must use it in collaboration with research institutions.
UBIQUITY: Ten years down the road, could you imagine someone like Steven Spielberg coming along with a proposal that would make sense as a purely commercial application?
CHI: I would want to know if the commercial Internet could support such an application. As commercial applications continue to proliferate and get more elaborate, the network that supports them will grow to accommodate them. I see the growth and capability for those kinds of applications taking place within the commodity Internet, rather than spilling over into Internet2.
UBIQUITY: What are your thoughts on Internet3?
CHI: It might develop for certain applications that cannot run very well on Internet2. For instance, maybe they require special protocols that don't work well with the routers that move the traffic around Internet2. If that were the case and the money were there to make it happen, I could visualize someone designing and implementing a network making use of entirely different routing algorithms tailored for a special set of applications. But that's all speculation. To be safe one always has to assume that something unforeseen will eventually happen.
UBIQUITY: Were you startled at any point by the development of the Internet?
CHI: Even though I participated in those early organizational meetings, I wasn't involved in the evolution of this technology in a significant way. I was a computing center director and my concern was mainframes and getting the grades out in time and supporting the students. The materialization of the Mosaic Web browser was, to me, absolutely startling. Considering where we were at that time, what a beautiful piece of work that was! I think that it, along with the underlying HTML protocol were probably the most revolutionary developments in this whole business.
UBIQUITY: Talk about your own history. How did you get involved?
CHI: I started my career as a physicist but I was always active at my campus at the University of Albany in advisory capacities of one sort or another as they had to do with computing. In the early 80's the director of the computing center resigned and the associate director was made acting director. I was asked to head up a search committee to find the new director. After a couple of months the acting director left and the president called and asked me to mind the shop for a while, which I did and I never went back. Ultimately, I became director of the computing center. Some years later an opportunity here at NYSERNet materialized. In 1999 I left the university and came to NYSERNet.
UBIQUITY: How do you view your role in NYSERNet? Are you in the trenches fighting off attacks or is it the good life?
CHI: It's a little of both. My title is Director of Network Technologies. Two threads report to me. One is network operations, which means supporting the research network and dealing with its trouble; the other is network development which means planning for the future.
UBIQUITY: What is happening in network development right now?
CHI: Our most immediate concern is that our contract for our current facilities, for the backbone, expires next April. We're in the midst of trying to decide what should come after that. Any decision like that is informed not only by budget but also by demand, by market conditions, by the technology, and also by our neighbors. There is very strong interest on the part of the Ontario Regional Provincial Network to interconnect at Buffalo so we can exchange traffic and provide each other with an emergency back up if some disaster occurs somewhere. We have a similar relationship developing with the Quebec Regional Network and finally, we've been spending a lot of time with an emerging organization called NEREN, the New England Research and Educational Network. They are interested in implementing an optical fiber network that would span southern New England over multiple paths in a manner similar NYSERNet. This would include institutions such as Harvard, MIT, U Mass at Amherst, University of Connecticut and University of Rhode Island. They have an interest in bringing their network westward to Albany and toward the southwest to New York City and attaching to NYSERNet at both those places, which gives them redundant paths for disaster recovery and of course, conversely, gives us redundant paths for the same purpose.
UBIQUITY: Let's move on to a different topic. What are you most interested in besides your job?
CHI: I moderate an Internet mailing list whose topic of interest is pipe organs. It was cited by the Chronicle of High Education some years ago as the premiere scholarly pipe organ list in the world. We have about 1500 subscribers from all over the world. The list deals with pipe organs, pipe organ technology, pipe organ performance, organists and organ music. We're rather strict about making subscribers stay on topic and not get on to long tirades about how awful the minister is or how they'll never play that kind of wedding again, which tended to happen in the past. We also have a smaller list dealing with harpsichords and similar keyboard instruments.
UBIQUITY: Do you play yourself?
CHI: I used to play the piano and while I was in college, I played the organ. About 20 years ago I had a carpentry accident that damaged my right hand and put an end to pianistic activities.
UBIQUITY: What do you do for a musical outlet?
CHI: My late wife bought me a cello, for which the right hand holds the bow which my right hand is still able to do without difficulty. That is one of my two musical outlets. The other one, which is totally embryonic, is a long-time desire to play the carillon [chromatically tuned bells controlled by a keyboard]. The keyboard is similar to a piano with white keys and the black keys. The difference is that the keyboard consists of levers about a foot or so long and maybe one inch in diameter, fairly well separated from each other. You play by striking the keys with your fists. The Albany City Hall has one of the finest carillons in the country but it is very infrequently played. I have been trying for more than a year without success to get access to that instrument -- there were security considerations: How did they know I wasn't going to blow up City Hall? The fact that I had well-established credentials at the University didn't seem to make much difference. But I finally was told that they will make the carillon available to me during limited evening hours. So, my newest adventure is going to be to learn to play that instrument.