Enhanced gridway metascheduler for heterogeneous grid using trust and reputation

Abstract

Grid computing allows sharing of unutilized heterogeneous processing resources that are ready to participate in a grid around the world Scheduler an important component of a Grid, either through a single or multiple middleware is used to make use of Grid for various computationally intensive applications Issues in Grid computing are many folded and innumerable namely Grid computations Trust and Reputations Statistical analyses on history performance of nodes middleware technologies proposal of fixed and variable value coefficients to terms in computing trust values etc Taking into account the intensities of processing for applications the availability of appropriate Grid resources are found the Grid scheduler actually does only matchmaking by default based on the available capabilities of the resources What is promised by Grids or agents of Grids may be different from what actually the resources provide to users in reality Thus terms such as trust or reputation even though social in nature need to be predicted with near certainty because schedulers as such do not take into consideration of trustworthiness of resources It depends upon history of success rates resource availability factor network characteristics hardware capacities of resources etc It is also understood that client efficiency that includes network efficiency has not yet been considered in the literature