Estimating Grid Reliability Using Bayesian Networks

Abstract

Grid computing has emerged as an important field for complex systems with need of sharing of resources in a large scale environment, wide area communication and multi institutional collaboration. Grid computing is used where there is a need of large amount of computational power which cannot be provided by a single computer. Both grid and distributed environments are different from each other in context of large scale resource sharing in grid. With the increase in the size of these systems the need to evaluate their reliability is increasing day by day. Hence various methods for estimating reliability are being developed. Bayesian Networks (BN) can also be used to estimate grid reliability. BN is used to depict the interactions between the various components of the distributed system and show the relationship between them. The BN depicts the probabilistic associations between the system components in a very simple and easy way. The relationship is represented using directed acyclic graph where nodes represent the variables and the links between each pair of nodes represent the causal relationships between the variables. So BN is the best method that can be used to evaluate the reliability of the grid services.