Background

When we watch the recent spectacular science fiction and fantasy thrillers,
we remain totally under the spell of the amazing scenes and the underlying
special effects that are being displayed. We watch the movies, applaud the
actions, and come home totally enchanted and wanting for more. Most of
the time we forget about the enormous amount of effort that is required to
produce such a spectacle. Even if we acknowledge that, we tend to ignore
the proverbial “work horses” or computers that generate such remarkable
special effects. We have got used to the special effects so much that we
take the computing power required to provide such visual spectacles for
granted. It takes billions and trillions of CPU cycles to create special effects
like those in Spiderman [1], Shrek [2], and other such visual treats.
To satisfy the ever increasing hunger for better special effects among
movie-goers, more and more complex animations are being developed
which are continuously raising the bar for computing power required.
Requirements of huge amounts of computing power are not only limited
to the field of rendering and animation. Scientists are analyzing terabytes
and petabytes of data to provide better weather forecasting [3], develop
more efficient models for detecting natural disasters, high energy
physics [4], and so on. By virtue of the hugely popular SETI@Home [5]
project, most of us are aware of the enormous computing power required
for searching extraterrestrial intelligence. The project allowed people to
download the SETI@Home software in their own machines and run the
program in the screen saver mode. Apart from fundamental research, computing
power is also required in huge quantities in the life sciences industry
for drug discovery [6]. Financial industries require huge amounts of
processing power to do risk calculations, credit analysis, and so on [7].
Manufacturing industries are not very far behind. Simulations of automobiles
based on complicated mathematical models [8] require enormous
computing power.
Similarly, EDA and Oil & Gas explorations also require
computing power to do more computations in a shorter time to satisfy the
ever increasing demands of the market.
Therefore huge computing power is required in several industries.
Now if we look at the computing resources available, we will find that the
laptops of today are perhaps as powerful as servers a decade ago. Moore’s
law, which states that computing power doubles every eighteen months, is
valid even today and will probably be true for the next five to six years.
With the advancements in the field of multi-core technologies, this growth
can be extended further [9]. Therefore computing power is increasing and
so is the demand. In this rat race, researchers have found an able ally in the
form of networking. Between 2001 and 2010, while processing power is
supposed to increase 60 times, networking capabilities is supposed to increase
by 4000 times. This means that at the same cost 4000 times the
same bandwidth will be available in 2010 as compared to 2001 [10].
Therefore the computing architectures developed a decade back would
probably require a rethink based on the technological progress in the fields
of computers and networks. Last decade saw the development of a field
called cluster computing [11] where the different computing resources are
connected together using a very high speed network like the Gigabit
Ethernet or more recently Infiniband [12].
In addition to the technological progress and the huge requirement of
computing power, enterprises have also undergone a radical shift in Information
Technology (IT) operations in the last few years. Enterprises are
now witnessing increasing collaboration and data sharing among the different
participating entities, resulting in the need and use of distributed resources
and computing. Another important element that has increased the
complexity of IT operations is the need for integration of different applications,
middleware developed in different platforms and by different vendors.
We are also seeing a spurt of mergers and acquisitions which require
integration of technologies across enterprises. Moreover, the enterprises
are outsourcing the nonessential elements of the IT infrastructure. The dual
pull of requiring more computing power and the integration of heterogeneous
components into the IT infrastructure has led to the development of
grid technologies. The technology is seeing a classical evolution pattern.
Initiated and started from the academic and the research community to fulfill
their needs, it is slowly being adopted by the enterprises, especially
those who have high computing needs like the life sciences, finance, and
manufacturing industries. However, the promise of grid computing goes
beyond that and the next few years should see a gradual adoption of the
grid as a natural choice among the other enterprises. However, the widespread
adoption of grid computing as an automatic choice in enterprises
depends upon the ability of the researchers and practitioners in reducing
the pitfalls that lie on the way. One such pitfall is security
we will briefly look at the evolution of grid computing, its benefits, and concerns in
the upcoming posts