Multiagent-based approach for scheduling meta-applications in heterogeneous grid environments

Abstract

In this paper, we present a multiagent scheduling framework for several classes of meta-applications (applications consisting of multiple schedulable components). The scheduling of multiple interacting components across multiple sites is a complex problem, especially when the available resources and specifically network capacity are assumed to vary between sites. This paper proposes a scheduling mechanism that uses routing indices and intelligent agents in each node, to perform global scheduling in a collaborative and coordinated manner. The scheduling mechanism was implemented in a grid simulation environment that allows the incorporation of intelligent agents. The behaviour of the proposed scheduling model was analyzed and compared with three other scheduling algorithms, in terms of selecting the best execution sites that is most appropriate for specific application characteristics.