Grid Computing: The Trend Of The Millenium

Abstract

A grid can be simply defined as a combination of different components which function collectively as a part of one large electrical or electronic circuit. The term Grid Computing can similarly be applied to a large number of computers which connect together to collectively solve a problem (which may be of scientific interest in most cases) of very high complexity and magnitude. The fundamental idea behind the making of any computer based grid is to utilize the idle time of processor cycles. Simply stated, a processor during the times it would stay idle would now team up with similar idle processors to tackle various complexities. The role each processor plays is very carefully defined and there is utmost transparency in the working of each processor/computer in a grid. This is called the Division of Labor in the smart world of intelligent computing. In lay man terms this is equivalent to a student and his group of friends collectively solving a single assignment which contains more than a single problem. The solution is trivial but the effort is collective. A grid computing environment may take many forms. It could be molded as a cluster based, distributed computing environment based or peer-to-peer system. A cluster based environment would see a central computer often called a cluster head distributing or maintaining a job schedule of the other computers in the grid. A distributed environment is seen often in the web environment. For example when a user requests a popular web page from the web server and if the web server is experiencing traffic congestion, then the user is re-routed to the same page on a different web server. The transition takes place so rapidly that the momentary delay due to server bottleneck problems is hardly felt. Peer to peer computing can be best explained through music download engines. If a user has a file he decides to share it via the web, other users needing the same file copy it through their music download engines. With great computing power comes great responsibility. Security is of utmost importance in a grid environment. Since a grid performs large computations, data is assumed to be available at every node in the processing cycle. This increases the risk of data manipulation in various forms. Also we have to keep in mind what happens to the data when a node fails. An ideal grid will have a small time of convergence and a low recovery time in case of a complete grid failure. By convergence, we mean that each and every processor node will have complete information about each and every other processor node in the grid. Recovery time is the time it takes for the grid to start from scratch after a major breakdown. To put things in the right perspective, a good grid based computing environment will have an intelligent grid administrator to monitor user logs and scheduled jobs and a good grid operating system which will be tailor made to suit the application of the grid. We propose a similarity between the OSI model and a Grid Model to elaborate the functions and utilities of a grid. We also try to propose a queuing theory for Grid Computing. There have been numerous previous comparisons and each is knowledgeable in its own right. But a similarity with a network model adds more weight since physically a grid is nothing but an interconnection, and interconnection can be best defined in relation to a computer network interconnection. How do users access networked computers, how are files shared and what are the levels of security are best explained through a networked computer system.