Improving the Performance of Resource Selection in Grid Environment Using Job Categorization

Abstract

 The grid deals with large scale and ever expanding environment which contains millions of users and resources. In grid computing user can submit their applications transparently, without having to consider the location. But in order to maximize the performance of grid, the grid location is a critical factor because if the application which is communication based and submitted to remote location will increase the communication overhead and degrade grid performance and if it is computation based and submitted to low bandwidth will again affects the performance of grid. So we proposed an approach which is working on two phases. In the first phase we calculate Communication to Computing Ratio which is useful to decide the appropriate grid site for scheduling tasks. If the task CCR value is high then tasks will be executed locally and if it is low then tasks will be executed remotely. In the second phase we categorized the selection parameters based upon the CCR value and apply ranking mechanism to select the best resource.