Probabilistic resource state estimation in networked environments: the case of computational grids

Abstract

Computational grids (CGs) are large scale, geographically distributed networks of computing resources spanning distinct management domains. The dynamic state of these resources is only observable with an uncertainty induced by various grid environmental factors. In order to address this uncertainty we explore the use of Markov chains to model the grid resource state dynamics. The simulation results show that the model provide a resource state prediction within a small margin of error.