Rival academics are competing to create a new, faster internet. Sean Dodson investigates
More internet news

Sean Dodson
Guardian

Thursday November 8, 2001

There is another internet - already operational - where users are receiving connections up to 100 times faster than people at home. It is a network so swift and so powerful its advocates are claiming it has already changed the way we will interact with the internet in the future.

This new internet is being developed in universities and research laboratories across the globe. And although its usage might be confined to academics, its benefits could spill over into the mainstream in only a few years.

Until now, the race to build the next generation of the internet has been dominated by the US, and by one project. Internet2 is a consortium of 180 universities backed by the National Science Foundation and the US Federal government.

But from December 1 - for the first time in the history of the net - Europe's academic and research communities will be able to hook up to a faster network, covering more people and more institutions than its rival in the US. Geant, the new pan-European network serving more than 3,000 of the continent's academic and research institutions, will operate in 32 countries. It has a budget of 200 million euros over the next four years - 80 million of which has come from the European Commission.

Each of the 32 member countries already has its own national research network - the Joint Academic Network (Janet) in the UK, for example. But now, with universities and research institutions increasingly collaborating across different borders, they need a high-performance network that crosses those borders, too.

This is what Geant provides: a pan-European network at such high speeds it completely dwarfs the internet as we know it today. The kind of speeds available on Geant or Internet2 would impress even the most up-to-date technology companies.

So, if a researcher on the Janet network wants to hook up with someone in Nice or Barcelona, they tap into Geant. Similarly, if a scientist in Cambridge wants to collaborate with one at the Massachusetts Institute of Technology, they tap into Internet2.

"The Janet network peers directly with Abeline, the structural backbone of Internet2," says Jeremy Sharpe of Ukerna, the organisation that runs Janet. "We manage transatlantic network capacity. So researchers in the UK that are collaborating with universities in the States can have a direct path through Janet and on to Abeline."

Geant and Internet2 are not separate from the physical network of fibre optic cables and telephone lines that serve today's commercial internet. And it will not immediately replace the net as we know it. What both networks do is buy connectivity on the open market from the telecommunications network operators, and then earmark it for research purposes only. They don't lay any new cables and they don't dig up the road.

Dante, an organisation established in Cambridge eight years ago, built the Geant network. When it goes live next month, Geant will be an entirely new network, but it is also the fifth generation of a series of networks built by them.

Eight years ago, Europe was, according to its general manager Dai Davies, the poor relation when it came to telecommunications. And consequently each of Geant's previous networks has lagged behind the US. But the subsequent liberalisation of the European telecom market and the investment of a number of operators in pan-European networks such as Colt, Level Three and Global Crossing has changed all that.

"I think the significant change is that people have been building international networks within Europe over the last five years," says Davies. "Whereas, historically, the model has been that each country built its own network and connected it to the other countries."

But the story of the next generation internet begins not in Europe, but in the US, a shade over five years ago. On October 1 1996, in the unglamorous surroundings of a basement in O'Hare Airport near Chicago, 36 research scientists and technologists met to establish a formal project to build a new internet.

For years the group had felt that the growth of e-business was beginning to threaten the experimental nature of the net. A new network was needed, they decided, where academics and scientists could work on "advanced network technologies," unimpeded by the growth of email and commercial ventures. They called their project Internet2.

"Today, the commercial internet is very focused on delivering a reliable set of services to paying customers," explains Greg Wood of Internet2. "But the internet itself came out of an experimental environment where researchers and students and scientists worked together to do new things with networks. After the commercialisation of the internet, these three communities felt there was a need to return to that."

Initially, Internet2 was aimed at improving connections between the different networks already held on university campuses. But very quickly, the project moved on. The researchers began to ask themselves: if a faster, more reliable academic network could be achieved, what experiments could be run on Internet2? Soon after, they began to realise that these new experiments could completely transform our perceptions of what the net could do.

"The internet at that speed is not just a quantitative difference," says Wood. "It's not just that you can browse the web or send email 100 times faster. It means you can do completely new things and you need a new mindset to do them."

For example, the famous telescopes on the top of Mauna Kea in Hawaii are now hooked up to Internet2. Astronomers are now able to control the telescopes from their desktops in the research labs, such is the speed of the connection and the reliability of Internet2.

This is partly because, from the beginning, Internet2 included technology giants such as IBM, Cisco and Qwest. And already the lessons they have learned on Internet2 are finding their way into the commercial internet.

But it is not all about networks. The experiments to be run on Geant and Internet2 are equally fascinating. And they are not confined to the research lab. Even companies such as McDonald's, Johnson & Johnson and Ford are keenly watching developments on the new networks. The fast food chain has already shown interest in the tele-immersion experiments being run on Internet2. The company envisioned fitting tele-immersion cubicles in its restaurants so people away from home - even in separate countries - could have dinner with their family.

The other thing about the new networks is that they are beginning to fundamentally change the way institutions interact with each other. The key thing about the new networks is that they promote collaboration. Today, institutions have to cooperate because of the sheer scale of some of the things that they do. The Human Genome project, for example, would not have been possible without large-scale collaboration. Likewise, the information coming out of Cern is now so rich in data that no one institution can cope with it.

"There is now a recognition that individual countries having their own science programs is much less relevant," says Davies. "The new networks are a catalyst encouraging European based cooperation in research."

Europe might, for the moment, have a faster network. But the US has a head start when it comes to applications for the new internet. Geant's main problem is that - unlike Internet2 - it operates in 32 different countries, each with very different telecommunications markets. Portugal and Greece for example, are still constrained by a lack of competition in the telecoms markets. Bulgaria, Romania and the Baltic states are also lagging behind. But the centre of the network - UK, Germany, France and the Benelux countries - should benefit from the fastest network in the world for years to come.

"Internet2 plans to offer 10 gigabit capacity by 2003," says Marine Chartois of Dante. "By that time I think we will already be looking at 40 gigabits per second. That covers a larger area, more people and a much more difficult environment."

Telecommunications networks are like rail networks. Building your network first - as we have learned in the UK - does not mean you end up with the best network. If Geant is successful next month, part of the reason for that will be because it built its network after Internet2.

From data gloves to terabytes

Tele-immersion
Tele-immersion is the technology that will allow people in different parts of the world to feel as if they are sharing the same physical space. Virtual reality is really about the creation of fantasy worlds. Conversely, tele-immersion is about connecting real places in real time, through the development of "tele-cublicles". Earlier this summer, successful tele-immersion experiments were carried out at UNC Chapel Hill and the University of Pennsylvania in Philadelphia. Jaron Lanier, who coined the term virtual reality and who built the world's first data glove in the early 1990s, led the experiments, using Internet2. Similar experiments include the use of haptics - remote feedback instruments that allow users to "feel" objects thousands of miles away.

HDTV
The University of Washington in Seattle has been sending High Definition Television (HDTV) over Internet 2. The images are as good as the best photographic plates. Even usable X-rays can be viewed. This could have profound implications. Video conferencing with TV-quality pictures is one application; remote surgery another. Experiments have included a violin master class. A seminar at the University of Oklahoma allowed teacher and pupil - separated by several hundred miles - to play in harmony.

The Grid
Anyone who has encountered the Seti@home experiment - the smart screen saver that uses your computer to search for extraterrestrial life when you are not using it - will have experienced distributed computing, where the work is shared out among a network of connected computers. Experiments being run on Geant and Internet2 want to take distributed computing to a new level. When the Very Large Hadron Collider near Geneva becomes operational in 2005, terabytes of data will start pouring out of Cern. There are simply not enough supercomputers in the world to deal with that amount of data. That is where the new networks come in. The idea is that computers held on the networks could be connected to share processing power and even hard disc space. They will be locked in to a grid - effectively creating one super computer.

À Comments to online.feedback@guardian.co.uk