A mathematical predictive model for an autonomic system to grid environments

Abstract

One of the most important aims of the Grid technology is using geographically distributed resources. Nevertheless, Grid environments have a great problem: the system management is very complex because of the large number of resources. Thus, improving the system performance is a hard task and it would be advisable to build an autonomic system in charge of the system management. The autonomic system must take decisions based on the analysis of the monitored data of the whole Grid trying to improve the system performance. These decisions should not only take into account the current conditions of the Grid but the predictions of the further future behaviour of the system too. In this sense, we propose a mathematical model to decide the optimal policy based on predictions made thanks to the known past behaviour of the system. This paper shows our model on the basis of the decision theory. © Springer-Verlag Berlin Heidelberg 2005.