Title: STATE INDEPENDENT RESOURCE MANAGEMENT FOR DISTRIBUTED GRIDS
Abstract: Advances in network technologies and computing resources have led to the possibility of deploying large scale computational systems, such as those following a Grid architecture (or related architectures). The scheduling problem is a significant issue in the Grid environment. In practice, a scheduling algorithm should consider multiple objectives. Typically, there are two kinds of objectives. The first is the performance of the system in terms of quantities related to the completion time of tasks, the second is the amount of state information required, which is often measured in terms of quantities such as communications costs. These two objectives are often in tension with one another. For example, gathering large amounts of state information can lead to low completion times. In this work, we introduce a scheduling algorithm which simultaneously addresses the objectives listed above namely, minimizing completion times, while requiring zero dynamic state information. Using simulation, we show the promising performance of our algorithm, and its robustness with respect to errors in parameter estimates. Keywords-Grid Scheduling; Scheduling Algorithms; Shadow Routing Approach