A Comprehensive Task Scheduling Algorithm in Grid

Abstract

Grid is evolving to a more efficient global computing infrastructure by introducing more and more task scheduling algorithms. With the grid expansion and users’ requirements in an intuitive way, a new robust multi-criteria scheduling algorithm is needed. Resource brokers should provide a more comprehensive solution for users. Multiple scheduling criteria addressed by the related grid research include execution time, the cost of running a task on a machine and reliability. The existing scheduling approaches are usually dedicated to single criterion or certain criterion pairs that require users to define one’s preferences. These requirements are often not feasible for users and these algorithms often cannot find the solutions that fully satisfy users’ various requirements. The grid scheduling problem is NP-hard in the traditional grid system. In this paper, we propose a flexible and general requirement specification method based on N-constraint, and we model the problem as an extension of the Multidimensional Multiple-Choice Knapsack Problem. Further more, we propose a multi-criterion scheduling heuristic called N-Variable Constraints Algorithm (NVCA) based on dynamic programming for the optimal solution, dedicated to the problem model defined by us. Finally, promising results demonstrate the effectiveness of the proposed method with comparison to other existing methods.