Science - The Internet

Network news

One of the shortcomings of the commercial Internet
is that certain activities, like natural, normal human
interaction, are impossible -- it's just too darn busy.
But what if you could lease your very own data pipeline?

DAVID AKIN reports

Globe and Mail
Saturday, December 1, 2001

At precisely 11:49 a.m. on Friday, Nov. 9, on a stage at the University of Montreal, Nathan Picklyk tucked his violin under his chin and began to play the first notes of the first movement from Andre Lidel's Duet in D Major for Violin and Viola.

About 11 milliseconds later, Monica Guenter, in a studio a few kilometres away at McGill University, joined in, playing her part on the viola.

Watching this concert proudly but nervously, at the side of the stage near Picklyk, was Jeremy Cooperstock, a McGill computer scientist who had organized the performance and, in doing so, successfully demonstrated the world's first performance by two musicians sharing an Internet connection rather than a stage.

Cooperstock's demonstration was a watershed event for the elite club of the world's computer network engineers. No one had ever before been able to demonstrate that, under the right conditions, it is possible for natural, normal human interaction to occur over the Internet.

The commercial Internet is useless for this kind of activity. There are too many hiccups and stutters produced by what is called network congestion -- too much information from too many people trying to squeeze through too tiny information pipes.

To solve this problem, Cooperstock built his own network -- a private Internet, if you will -- between the Šcole des Hautes Študes Commerciales at the University of Montreal and his studio a few kilometres away at McGill.

While the high-speed residential Internet services usually run at speeds of about one megabit per second and many colleges and universities in Canada use a 10-megabit-per-second connection for their entire campus, his private network of fibre-optic cables was capable of moving 224 megabits each second.

They say "the medium is the message. Well, today, the medium must be faster than the message. Latency is the interaction killer," Cooperstock said.

His accomplishment was figuring out a way to move full-motion, DVD-quality audio and video through several computers, TV monitors, video cameras and kilometres of fibre-optic cable in the same time it would have taken a sound wave to travel from one of a concert stage to the other.

In fact, that was his benchmark. If Guenter and Picklyk had been standing on either ends of a stage, it would have taken 15 milliseconds or 15 one-thousandths of a second for sound to travel between them.

And while Cooperstock acknowledges that his test, done under near-perfect controlled conditions, is a long way off from real-world conditions, the demonstration pointed to a future where high-capacity Internet networks can transmit immense amounts of data quickly enough that they can help in the creation of virtual-reality environments.

These environments would let human beings, separated by hundreds of kilometres, carry on conversations, conduct master classes, collaborate together and play with each other as if they were standing a few metres apart.

Still, despite Cooperstock's groundbreaking work, this brave new world will be impossible given the current economics of providing pristine high-speed Internet connections.

His experiment was part of a $1.5-million research project. The Internet connection he used would probably cost hundreds of thousands of dollars a month, if you could get such a connection at all.

But Bill St. Arnaud thinks he has a solution.

St. Arnaud, senior director for advanced networks at the Ottawa-based non-profit group CANARIE Inc., was among the hundred or so scientists who witnessed Cooperstock's demonstration and he is convinced that the network power used by Cooperstock can, within a few years, be used by all Canadians, even those in remote and rural areas.

His idea is to let Internet users rent the coloured pulses of laser light that carry the world's Internet traffic inside fibre-optic cables.

So, rather than pay a telecommunications company tens of thousands of dollars a month for super-high-speed network services, a school would just buy a long-term lease for a beam or two of light capable of carrying gigabytes of data.

St. Arnaud and his colleagues at CANARIE estimate that, after a relatively small one-time capital cost to buy the computer equipment that gets hooked to the end of a fibre-optic cable, the monthly bill for nearly limitless bandwidth could be hundreds of dollars a month.

"I consider Bill St. Arnaud to be one of the great visionaries of the networking world," said Brian Reid, a professor of the practice of computer systems at Carnegie Mellon University West in Palo Alto, Calif. "His vision, and his dedication are extraordinary. When the history books are written to show how the world got changed by being connected, Bill St. Arnaud is going to rate a chapter."

St. Arnaud, born to a French father and an English mother in British Columbia, grew up in Ontario. He speaks only English. His brother, a PÈquiste, is bilingual, but he insists on speaking only French and lives in Quebec City. Yet another brother is happy to speak either official language and works for the federal government in Ottawa. "We're a microcosm of Canada in one family," St. Arnaud jokes.

Like many of those who helped to develop the Internet and computer networks more generally, St. Arnaud picked up what he needed to know about computers and networks by fiddling around with them. After finishing an engineering degree at Carleton University in Ottawa, he ended up working in Toronto, where he eventually started his own computer-networking business.

In 1993, he joined CANARIE, which was created with federal and private sector funding to help to develop advanced high-speed computer networks in Canada.

By 1998, CANARIE, with the help of companies like Nortel Networks, Bell Canada, and Cisco Systems, was ready to build its own high-speed research network, CA*Net, which let St. Arnaud try out his ideas about cheaply setting up and running an advanced network.

CA*Net soon became the network backbone that links the country's provincial and regional networks. Indeed, the network that Cooperstock used for his demonstration is part of Quebec's high-speed research network, the RÈseau d'informations scientifiques du QuÈbec.

CA*Net is powerful enough that if a researcher hooked up to the RISQ network in Montreal wanted to send the entire contents of the Library of Congress to a colleague at Simon Fraser University in Burnaby, B.C., it would take less than a second.

But funding for CA*Net runs out in July and CANARIE has petitioned the federal Liberals for $120-million to continue running and improving the network. There is some concern that Finance Minister Paul Martin may block funding for CANARIE's project because Industry Canada -- whose minister is Martin's leadership rival Brian Tobin -- is responsible for supervising the organization.

In fact, some network engineers in Canada and in other parts of the world worry that CA*Net may be this decade's Avro Arrow, a Canadian technology project that is admired around the world but died for political reasons.

"Most of the folk I know regard CA*Net as the premier effort in the world," said Gordon Cook, a New Jersey-based computer-networking expert who publishes the industry newsletter The Cook Report on the Internet.

St. Arnaud is happy to have the accolades, although he and his colleagues at CANARIE will be happier still to win funding approval from the federal government.

"I guess what motivates us is we're changing the course of history," he said. "We're not here for the money. We're leading the world and we're having an impact that the world is recognizing and that is a powerful motivating thing. Everybody's coming to us and saying, 'Wow! That's neat what you've done!' That's very gratifying.' "

In St. Arnaud's view, it should be possible to buy just one wavelength, or colour, of laser light that runs through one of the dozens of strands of glass that make up piece of fibre-optic cable. He calls his concept condominium wavelengths. Each strand can carry hundreds -- some researchers say thousands -- of wavelengths of light.

Cooperstock is ready for St. Arnaud's future. For his work, which requires immense amounts of bandwidth, he could lease a wavelength as easily as making a telephone call. He would -- if the research St. Arnaud hopes to do works out -- punch in a series of numbers into a Web browser on his computer to set up a high-speed line to a studio in L.A. and then, once he has done that, he could set up a link with an orchestra in Paris.

"Now, this is not going to happen overnight," St. Arnaud said, "but what we want to do is give a handful of wavelengths to [researchers] and say: 'These are your wavelengths, do whatever you want with them. You don't have to ever tell us.'

"Science can be carried out anywhere. That means science can be carried out as easily on a university campus as well as a school or, eventually, a home. Now everybody can participate in research."

The idea of taking science out of the lab and putting it on any computer anywhere, using high-speed research networks, is not just a pipe dream. As active as CANARIE has been in developing these advanced networks, it has been just as active encouraging development of things to do on those networks.

So, for example, high-school students in Canada and the United States are using high-speed research networks to run the world's largest cosmic-ray detector, under the supervision of the Centre for Subatomic Research at the University of Alberta in Edmonton.

Students and their teachers have installed cosmic-ray-detector dishes on the roofs of about 50 schools across the continent. Each dish is wired into a computer at the school that, in turn, is connected to every other computer participating in the project as well as the main computer at the University of Alberta. That main computer is then linked, via advanced research networks such as CA*Net, to satellites and other research centres around the world.

"The students are so enthusiastic because now they're not being spoon-fed science, they're participating in real research," St. Arnaud said.

CANARIE's researchers have had some preliminary discussions with Environment Canada to move weather modelling and forecasting off of supercomputers -- analyzing weather dynamics takes immense amounts of computer horsepower -- and let all Canadians participate in forecasting the weather using nothing more than a standard desktop computer that is hooked into a broader research network.

Right now, Environment Canada's weather forecasting supercomputers, located in Montreal, produce forecasts accurate to about a 50-kilometre grid. But researchers are now working on a way for home or small-business users to take that data from Environment Canada's supercomputers, combine it with weather data gathered locally, and produce weather forecasts accurate to within about one kilometre -- to the delight of golf course operators, ski hill owners, farmers, outdoor enthusiasts and sports teams.

Meanwhile, at McGill University this fall, Cooperstock is trying to bring this future into the present as quickly as possible. He has a 100-megabit-per-second Internet connection installed at his home and he plans to be a living example of the worth of such high-speed connections.

Among other initiatives, he will be conducting his classes from his home via his Internet connection. The course he will teach this way? "Human-computer interaction, strangely enough."