Research On Scheduling Algorithm For Independent Task Of Grid

Task scheduling is a fundamental issue in achieving high performance of grid computing. Scheduling strategies directly effect the performance of grid. The features of distribution, heterogeneousness, dynamic and self-ruling of grid challenge the traditional scheduling algorithm. Task scheduling algorithms is one of the most popular topics for the Grid technologies research.This paper firstly introduced the conceptual background,category and system structure of Grid. Then discussed the goal, principle and basic strategies of task scheduling in Grid computing. Among the seven popular Grid task scheduling algorithms, the paper emphatically analyzed the shortage and limitation of Min-Min and Max-Min algorithms from two aspects regarding task's equality and algorithms'performance. Based upon the Grid heterogeneous, the paper aimed to load balance of task scheduling, and it provided a Grid task scheduling algorithm for Fast- Balance task in heterogeneous system. This algorithm adopted the greed strategy. At first it dispatched tasks by unit to every machine accordingly. Then it found the maximal loaded machine and the minimal loaded machine. And this transferred or exchanged those tasks on the two machines that make tasks fast balance. According to iterative operation, it made every machine's tasks fast balance in Grid system. At first theory demonstrated algorithms'accuracy, then the paper did many simulation experiments using Grid scheduling simulation toolkit (GridSim) on JCreator IDE environment that makes Min-Min and Max-Min algorithms as criterion. As the results of experiment, the Fast-Balance algorithms provided a very good performance and can get a better result than Min-Min and Max-Min algorithm.At last, the dissertation made a conclusion with the summarization of this research and indicated the direction of its future research.