Predicting the reliability of resources in computational grid

The computational grid built on wide-area distributed computing systems is a more variable and unreliable computing environment, hence it is undoubtedly necessary to predict the reliability of the resources before allocating them to a grid application in order to ensure certain availability for the application. In this paper, we present an algorithm for predicting the reliability of resources based on time series of performance data and status data generated by the Grid Monitoring Service. The predicted reliability information can augment Grid Information Services. Simulation results show that with the reliability information, the grid resource allocation service can guarantee the high availability of the applications in the computational grid.