Print

Print


hen the Swiss-based pharmaceutical giant Novartis needed a new 
supercomputer for designing drugs, the company found it already had 
one. It was hidden in the unused computing power the company had 
available in the thousands of PC's that were already being used in its 
offices.

Novartis used American software technology to harness the power of its 
office personal computers, but European and American scientists and 
government officials said that Europe was moving faster than the United 
States to capitalize on the approach, which is called grid computing.

Grids lash individual computers together, tapping their unused power to 
tackle complex computing chores beyond the scope of isolated 
processors. They are being called upon by scientists and corporations 
for a variety of applications, including building low cost 
supercomputers and creating work groups that can span cities or even 
the globe.

The shift underscores more than a new style of computing. It also 
signals a new reality in the transfer of computer technologies. The 
global Internet is accelerating the rate at which new technologies can 
be deployed anywhere, frequently shortening or even erasing the 
certainty of American technological leadership.

"Europe has decided that this is a real competitive advantage,'' said 
Peter A. Freeman, assistant director of the National Science Foundation 
here for computer and information science and engineering. "And they 
are going after it."

Novartis used software by United Devices of Austin, Tex., to link 2,700 
desktop personal computers to help create drugs. This summer the 
company said that it had discovered several promising new chemical 
molecules with its grid and it planned to expand the system to its 
entire corporate network of 70,000 personal computers.

Europe's rush to grids underscores cultural and political differences 
between it and the United States, technologists in each area say. While 
American universities and companies often lead the innovation parade, 
the United States sometimes becomes hamstrung in putting new 
technologies into use by a proliferation of competing computing and 
telecommunications standards and by government reluctance to 
orchestrate industrial policies.

By contrast, European governments have traditionally been more 
effective in deploying unified standards and concentrating on 
technologies that appear to offer an economic advantage. But that extra 
help can lead to big mistakes, occasionally pushing technologies so far 
ahead of the market that they never deliver a reasonable payoff.

Cellphone networks are an example of the difference. Although the 
technology was invented in the United States, the current European 
digital cellular networks are generally acknowledged to offer superior 
service. But Europe's telecom companies have wasted tens of billions of 
dollars buying the rights to deliver third generation, or 3-G, 
cellphone services that have generated little interest. With grid 
computing, Europe may have as much as an 18-month lead in deploying the 
advances in practical ways, European scientists and government 
officials said.

While the United States is beginning to respond to a report in February 
from the National Science Foundation Advisory Panel on 
Cyberinfrastructure urging coordinated investment in grid technologies, 
the European Union is preparing to start two major initiatives in early 
2004.

One, called Enabling Grids for E-science in Europe, aims to build the 
largest international grid infrastructure to date, operating in more 
than 70 institutions throughout Europe, providing 24-hour grid service 
and a computing capacity comparable to 20,000 of today's most powerful 
personal computers.

The other is a distributed supercomputing project, led by France's 
National Center for Scientific Research, that will connect seven 
supercomputers in Europe at optical network speeds, getting a leg up on 
the TeraGrid project in the United States, which aims to connect the 
nation's major supercomputer sites.

"The goal is to establish Europe as one of the most dynamic and 
creative environment in the world to deploy grid-enabled 
infrastructures," said Mário Campolargo, director for the research 
infrastructure unit at the European Commission in Brussels.

The strategy appears to be leading toward accelerating the deployment 
of commercial projects like the Novartis grid.

There are a "number of research-oriented organizations in Europe that 
have made significant early progress," said Andy Butler, a vice 
president for enterprise systems in the London office of the Gartner 
Group, a technology research firm. Europe's advances, he added, mean 
"vendors like I.B.M., Sun and Hewlett-Packard have made a lot of their 
early progress in Europe, as opposed to the U.S."

The Europeans also have the advantage of a clearer road map than in the 
United States, where planning for computing and networking 
infrastructure is scattered throughout the federal government.

"The European Union has a 5- to 10-year strategic plan finalized,'' 
said Larry Smarr, a grid computing pioneer who runs an institute 
associated with the San Diego and Irvine campuses of the University of 
California. "This is a slap in the face and a wake-up call that things 
have gone global.''

The Hewlett-Packard Company, for example, said last week that it had 
joined BAE Systems, a British maker of aerospace and defense systems; 
Cardiff University; the University of Wales, Swansea; and the Institute 
of High Performance Computing in Singapore to use grid computing for 
advanced, collaborative simulation and visualization in aerospace and 
defense design.

The project is being paid for in part by Britain's Department of Trade 
and Industry and will try to solve computer security problems when 
using a grid because businesses like BAE need to control what 
information should be protected from outsiders.

When it comes to grid projects "geared towards getting real 
applications running, there are probably more in Europe," said Martin 
Walker, an executive at Hewlett-Packard involved in technical computing 
for Europe, the Middle East and Africa.

The British government is helping to lead that drive: It is supporting 
a variety of projects, including the Diagnostic Mammography National 
Database project, which received matching funds from I.B.M. and aims to 
use grid computing to create a new model for scanning, storing and 
analyzing mammograms.

"The nice thing about'' the project "is that it is linked to the 
government's e-science program, not just a random collection of work," 
said Brian Carpenter, an engineer in the storage and networking section 
of the I.B.M. Systems Group. The link with the government means "there 
is a very good chance that it will be integrated into the national 
health care system," he said.

The British government alone will spend $335 million on grid computing 
from 2000 to 2005, said John M. Taylor, director general of research 
councils in the Britain's Office of Science and Technology and a former 
director of Hewlett-Packard Laboratories Bristol, the European arm of 
the company's long-range research laboratories.

"The technology is mostly still coming from the U.S.," said Ian Foster, 
associate division director for the distributed systems lab at Argonne 
National Laboratory in Argonne, Ill. "What is happening now - which is 
either good or worrisome, depending on your perspective - is that there 
is a tremendous amount of investment in the European Union."

Beyond money from individual governments, the European Union is 
expected to spend $428 million from 2002 through 2006 to upgrade the 
grid's infrastructure. Unlike in the United States, all the projects 
have fixed objectives and private sector partners.

Grid backers argue that lagging American planning and financial support 
is an issue in part because of evidence that the creation of computing 
grids will have a direct economic impact. Last month, a study released 
by the Rural Internet Access Authority, a North Carolina economic 
development group with industry backing, estimated that deployment of 
an advanced computing grid in the state would add more than $10 billion 
and 24,000 jobs through 2010.

The United States is ahead on one front: it has made the most progress 
in the deployment of computing grids for scientific applications like 
studying earthquake risks. Next year, the TeraGrid project is expected 
to offer computing speeds of up to 20 trillion mathematical operations 
a second and the ability to store a petabyte of information - about 
what could be saved on 25,000 standard personal computer hard drives.

Europe's equivalent effort, Openlab, involving I.B.M. and a research 
center in Switzerland for the Geneva-based European Organization for 
Nuclear Research, known by its French acronym, CERN, is not expected to 
reach the same level until 2005.

But the Europeans are racing ahead in developing faster optical 
networks. A CERN-Caltech team set an Internet 2 Land Speed Record 
recently by transferring 1.1 trillion bytes of data in less than 30 
minutes.

Such transfer speeds were "not even thinkable a year ago," said Flavia 
Donno, the computer scientist in charge of physics experiments on the 
grid for CERN.

Now the Europeans are ready to move to 40 gigabits a second, relying in 
part on what is known as dark fiber, unused high speed fiber optic 
infrastructure. The test beds will allow scientists and businesses to 
share information and computer infrastructure in real time.

The work being done in Europe is much more concentrated on building 
something that is ready for end-users than are the grid projects in the 
United States, said Dr. Donno, who worked with American scientists to 
build the first interoperable grid between Europe and the United 
States. With Europe ahead in getting large applications running, 
American scientists have asked for a role in the Enabling Grids for 
E-science in Europe project. But because European research and 
development programs do not normally provide financial support for 
American participation, the Europeans are asking the National Science 
Foundation to contribute to the project. Mr. Freeman, the foundation 
official, met recently with European Commission officials on the issue.

"I hope they will find an agreement that is a good balance between 
national competitiveness and international cooperation," said Fabrizio 
Gagliardi, project leader of the European DataGrid middleware project. 
There is "so much to be gained if we join forces,'' he said. "We need a 
worldwide infrastructure so that we can really work to solve basic 
problems like the energy crisis and sustainable development."

Copyright 2003 The New York Times Company | Home | Privacy Policy | 
Search | Corrections | Help | Back to Top
-- 
Kelvin Chu, Physics Department, Cook Building
82 University Place, University of Vermont, Burlington, VT 05405-0125
http://www.uvm.edu/~kchu/; (802) 656-0064; Fax: (802) 656-0817