The science grid
Click here for more photos.
Scattered across the globe is a wide range of unique
scientific resources: large experimental facilities (which
will include the Spallation Neutron Source under
construction at ORNL), supercomputer centers, petabyte
data archives, the high-speed Internet, and, most
importantly, the expertise of scientists at laboratories and
universities around the world.
"The easy science problems have been solved," says Thomas Zacharia, ORNL's associate
laboratory director for Computing and Computational Sciences. "Tomorrow'ís scientific
breakthroughs in biology, nanotechnology, and physics now require large,
multidisciplinary teams and the ability to effectively use the scientific resources
distributed around the country." In fact, all scientific fields have the need to bring
resources to the fingertips of area experts.
To connect scientists, instruments, computing, and data, the Department of Energy is
establishing a DOE Science Grid, one of several grids being constructed by science
agencies, including the National Science Foundation and the National Aeronautics and
Space Administration. According to the DOE Science Grid Web site, "The vision for
'grids' is to revolutionize the use of computing in science by making the construction and
use of large-scale systems of diverse resources as easy as using today's desktop
environments." Collectively, all these grids are part of a global science grid.
This science grid creates a "virtual laboratory" environment that allows scientists to
tackle tomorrow's problems more effectively. It enables innovative approaches to
scientific computing through secure remote access to online facilities, high-speed Internet
access, collaborations among scientists separated by distance (e.g., at different DOE labs
in "collaboratories"), shared petabyte datasets, and large-scale distributed computation.
Scientific communities must have the capability for multiple users to remotely access
high-performance computing resources and large data archives to both perform
simulations and analyze the results of experiments. Researchers also need to collaborate
with scientists at different sites who are involved in the simulations or experiments. They
must coordinate access to and use of the resources.
The complex and evolving nature of scientific discovery requires general services that
can be combined in many different ways to support different problem-solving approaches
and the ability to evolve along with the scientific understanding of the problem. Resource
management for such dynamic and distributed environments requires global naming and
authorization services, scalability, fault tolerance, data management, security,
authentication, and protection of proprietary data.
"The goal of the science grid," says Al Geist, who is overseeing ORNL's efforts in
developing the DOE grid, "is to provide a common and supported set of services across
all the scientific resources so that scientists can easily access, use, and share these
resources and results more efficiently with the larger scientific community."
The design and deployment of large, multi-site grids are still evolving. You can learn
about the state of the art and stay abreast of new developments by accessing this ORNL