A supercomputing grid in the USA which will help scientists tackle
highly complex problems is likely to be available in December, two
years ahead of schedule.
IBM and the US Department of Energy's (DOE) National Energy
Research Scientific Computing Center (NERSC) will collaborate on
creating a DOE Science Grid which links supercomputers in several
locations around the USA.
The Science Grid will enable scientists at US national laboratories
and universities around the country to perform more complex
calculations and computer modelling, and manage and analyse large
datasets, IBM said. The Science Grid will eventually have a
capacity of 10 trillion calculations per second and storage
capacity equivalent to 200 times the number of books in the Library
of Congress, IBM said.
Grid computer systems allow organisations to share applications,
data and computing resources over the Internet, using protocols
provided by the Globus open source community. Typical large-scale
projects which require the use of supercomputing grids include
global climate, genomic and astrophysics research.
NERSC runs a 3,328-processor IBM supercomputer, which will
eventually link via the Internet to supercomputers, data storage
and experimental facilities at Lawrence Berkeley, Argonne, Oak
Ridge and Pacific Northwest national laboratories.
The Science Grid will make extensive use of open-source software,
including the Linux operating system and Globus. The Globus toolkit
includes grid computing tools and libraries for security,
communication, information infrastructure, fault detection,
resource management portability and data management.
A core group of NERSC's 2,100 users will be able to begin accessing
resources via the DOE Science Grid by the end of the year.