| |
The metacomputing project
Whatis a metacomputer
The Legion Project
The Milan Project
|
|
 |
 |
What is Metacomputing?
Metacomputing involves using multiple computers linked by
high-speed networks to solve application problems; they appear to
the user as a single computer but provide a level of performance
far in excess of individual computers in the network.
Metacomputers or computational grids function like a networked
virtual supercomputer. Metacomputing was born out of a need to
utilize greater processing power than was available in a single
site, and to combine the power of computers with different
architecture.
Metacomputing is also useful for collecting, manipulating and
analyzing data from remote databases and instruments like
microscopes, telescopes and satellite downlinks. Large-scale data
intensive applications require high-performance computing,
high-speed networking, data storage facilities and interactive
software.
In addition to science and engineering, high-performance computing
is now required for creating new films, weather forecasting,
designing new drugs, etc., and system simulation is used by
engineering companies for rapid prototyping and to reduce the time
to market for new products.
Metacomputing provides users with the power of a supercomputer,
without the high cost. Small companies need high-performance
computing for developing new products, but their limited
requirement may not justify the high installation and maintenance
costs of such systems.
Metacomputing offers an economical solution and allows small
companies to access high-performance computing capabilities, as
and when they are required. Users can pay for their actual usage
and avoid the burden of high ownership costs.
Metacomputing allows users to gain access to desktop
supercomputing with powerful graphics capabilities and to use
distributed supercomputing to solve complex problems. People can
interact and collaborate with other users in different
geographical locations.
When high performance computing facilities are centralized there
may be a lack of flexibility and redundancy. A failure could cause
major disruption to the business.
Distributed heterogeneous computing uses computing installations
in different locations, linked by a high-speed network. There is a
big reduction in the requirement of bandwidth and a failure in one
component is a local problem and does not cause the collapse of
entire grid. Different service providers can be a part of the
network and compete for customers, leading to a reduction in
prices and enhanced service quality.
|
|
Globus
Other Distributed Computing Links |
