Preamble
We have nearly 400 million PCs in the world. On average each PC is used for 10%
of the time. Remaining time all the PCs are idling. This is one scenario of under
utilization of available computer power. On the other hand, researchers who want
to solve a gigantic problem like stellar physics of several petabytes data are in
need of large computer power which is not available to them individually. It is
beyond the storage capacity of any desk top computer and usually requires some 25
tera-flops (floating-point operations) per second of computer power.
Today the fastest super computer can handle a maximum of 4 Tera-flops/second. In
the last 5 years, we could increase the raw speed of individual computers by a factor
of around 1 million but still no where sufficient to handle many challenging scientific
computing problems. This is second scenario where people are in need of large computer
power in this work. Suppose we create an environment where we can cluster all the
computers in the world and create wide area parallel and a distributed computing
with necessary networking protocols. Such an infrastructure called ‘Computer Power
Grid’ (CPG) will enable sharing, selection and aggregation of a wide variety of
geographically distributed computational resources including storage systems, data
sources etc owned by different organizations for solving large scale resource intensive
problems by researchers in science, engineering and commerce.
CPG will be the future research area which will shortly evolve into a multi billion
industry like electric power grid industry with impact on every aspect of human
living.
What is Computer Power Grid?
Grid computing is a large scale networking methodology of distributed computing
and virtualization of data resources such as processing, network bandwidth and storage
capacity. CPG grants its users seamless access to vast IT capabilities as though
interacting with a single large virtual computer. Such technology will influence
the way we do research and transform scientific, engineering, commerce and many
more disciplines.
Brief History
In 80’s, one could link two computers through ‘internet working protocol’. In 90’s,
hyper text protocol came into existence; WWW (World Wide Web) exploded enabling
linking of any two documents very convenient. In 1995, concept of grid computing
was introduced by Foster and Kesselman & Stevan Tuecke while working on protocols
and standards. They continued to develop this concept and for the first time, instrumental
in integrating peer to peer computing and web-services to provide seamless access
to remote mammoth computer power. Virtually, their protocols provide any computer
the ability to peep into cyber space irrespective of location and avail resources
from any nook and corner and use them for any power hungry application. With realization
of CPG, organizations can optimize computing and data resources and pool them and
share them judiciously with due regard to commercial viability.
What is challenge in CPG?
We know the resources which are to be put into Grid are geographically distributed
across multiple administrative domains with varying availability, heterogeneity,
diversity of use, varying cost and varying adaptability. Managing such vast and
variable resources and scheduling of purpose at viable cost constitute a complex
task. Basically CPG involves three collaborative parties namely the resource owners,
resource customers and resource distributors who can join hands based on economic
based frame work with certain trade-offs. Each party has to work within certain
demand-supply working model for best win-win results.
Examples of large internet computing projects
and software tools
– Computer Grids
– SET1 @ home internet computing
– Parabon Computation Inc.'s Compute Against Cancer project, which analyzes patient
responses to chemotherapy
– Entropia Inc.'s Fight AIDS at Home project, which evaluates prospective targets
for drug discovery.
– IBM-GMAS: Grid Medical Archive solution for advanced storage technologies for
health care.
– US National Science Foundation’s National Technology Grid (NSG Tera Grid)
– NASA’s information power Grid 3 (NASA IPG) World Wide Grid (WWG)
– European Data Grid
– Particle Physics Data Grid
– Grid Physics Network (Gi Phy N)
– Network for earthquake engineering simulation Grid
– Nimrod-G for providing algorithm for scheduling parameter sweep applications
– Grid-sim for simulating grid environment.
– Globus Tool Kit developed by Argonne National Laboratory and University of South
California’s information Science Institute (Carl Kesselman’s Team)
Examples of existing software tools
– GRACE (Grid Architecture for Computational Economy): Economy based grid resource
management and scheduling system.
Future areas of research in CPG
It is a growing field, but still at its infancy. It offers a large scope for
prospective researchers to undertake research irrespective their basic discipline.
Some topics are mentioned below, though the list is not exhaustive.
– Development of standard protocols and software tools to handle such computer power
grids.
– Development of open-architecture and open-source infrastructure needed to construct
CP Grids.
– Grid simulation and adaptive grid generation techniques.
– Management of grid resources through scalable, controllable, measurable and easily
enforceable processes with regard to scalable network architecture.
– For a given deadline and budget, GC applications scheduling with cost and time
optimization strategies.
– Standards based grid portal
– Management of computer grids and impact on society and human behavior.
– Modeling of computer power grid as an industry with regard to economics, cost
of maintenance, trading and quality of service.
– Relationship model between CPG resource owner, provider or distributor and consumer
akin to any business venture like power grid.
Conclusion
This 21st Century technology is going to change the way we live. CPG which
links various computing resources will enhance human communication; provide access
to unavailable data, large computation of economic scale amongst others. It will
provide opportunity to join hands between CPG provider companies with scientific
community for better human welfare. Future research will be carried out in virtual
laboratories in which researchers will work in collaboration without regard to their
physical location as well as resources location at much lower research cost.
Today we need CPG to solve many mega unresolved data handling problems in biophysics,
chemistry, biology, scientific instrumentation, drug design, tomography, high energy
physics, data mining, financial analysis, nuclear simulations, material science,
chemical engineering, environmental studies, climate modeling, weather prediction,
molecular biology and neuroscience. This list is not exhaustive.