Computing consists of a network of heterogeneous computers, from which a virtual super computer is essentially formed. It displays immense potential as the various resources across large networks can be pooled to service many and be utilized by many, using the Internet from around the world. The potential for parallel CPU processing is one of the most attractive features of a grid. A perfectly scalable application will finish five times faster if it uses five times the number of processors. Application software as required by the users of the grid. Thus, the structure can be represented in layers, as implied by the grouping of grid components. Hardware, the bottom layer, would then contain a large number of heterogeneous resources and would be accessed by a limited number of users to ensure data privacy. The next layer would then consist of application software and tools that are useful for the users and which are domain-specific. In this research we analyzed the distributed and high-performance system in grid computing to provide the efficient resource discovery and message broadcasting. The Distributed Spanning Tree (DST’s) implementation is altered and adapted to achieve better server load and message load distribution by a selective search and look-up mechanism in this proposal. In addition, a fault tolerance mechanism is also expressed in this contribution, as part of the DST’s adaptation, such that if the system which is providing the service fails or leaves the grid environment, then the backup site will immediately take up the execution and recover the task.
展开▼
