ABSTRACT Computers are now ubiquitous. Frequently, the power of these computers is underused,
but rather than harnessing their unused capacity to carry out our business or leisure activities, we buy additional, more powerful machines. When computers started evolving, almost all IT organizations were in assumption that by adding more CPU’s, memory and disk to large data processing servers we can process large data volume in considerable amount of time. This assumption however came at a higher cost and IT organizations felt the need to find the alternative solution which will be more cost-effective and solve the problem of extensive data processing. In early 1990s, a new
technology was evolved which is known as ‘Grid
Computing’ and it is becoming a rapidly emerging field which has promised to change the way we tackle complex problem. A grid is typically a collection of low-cost servers connected over a high-speed network in which resources are pooled and shared. The idea of ‘grid’ is to enable local or distributed networks of computers to be linked together, providing an opportunity to harness the full potential of the various components. Although this is a relatively simple concept, achieving it has been a major technological breakthrough. The Grid Computing technology is widely used now-a-days in most of the IT companies like IBM, Informatica, SAS, etc. NASA has effectively started building the grid-enabled applications grids to maintain and store volumes of data used in projects examining the Earth, Solar system and Universe. The data collected by space machines in NASA is represented in large matrices of 1000x1000. So the calculation of such large matrices might overload the most advanced computing systems at NASA if they don’t use the Grid. The article on ‘Grid computing’ discussed here has the personal project experience in which grid technology was implemented over client-server architecture. The project comprised of implementation of an algorithm, software and a client-specific screensaver. An algorithm was developed for resource brokering which will break the computational task of server into parts based on actual job requirement and client machines availability. A screensaver was developed which was installed on all client machine. Whenever a client machine is free, this screensaver will start and the notification will be send to server about client machines’ availability. Like this, all the idle client’s availability will be notified to server machine so that it can utilize the CPU time of idle client machines for his computational task. The algorithm then breaks the computational task of server machine and divides the task among the available client machines. The task will be processed individually at each client machine and each will return with the result of computation to the server. In between, if client machine comes out of the idle mode then it will notify the server and return the intermediate result. This way the CPU time and other resources of an idle client machines is used for performing the task involving large and intense computations. By implementing grid in client-server architecture one can effectively use its available resource power without increasing the overall cost.