The Research Of Comprehensive Evaluation Decision Model In Grid Resource Allocation

With the development of computer technology, grid computing becomes research hotspot gradually in recent years . Grid integrates high-speed Internet, computer, large-scale data bank, sensor, remote equipment, to realize the work of comprehensive sharing and cooperation. In grid environment, the target of resource selection and task scheduling is satisfying reasonable matching in application program and calculating resource, to improve the execution efficiency of grid application. Research organization carries out research from economic view at home and abroad at present, solves the problem of grid resource allocation by market mechanism. In wide-area and heterogeneous grid environment, grid performance correlates with the heterogeneous nature of resource, network performance and reliability . so grid environment must consider many factors and evaluation indexes to guarantee service quality . Based on the above requirement, this paper proposed multiple attributes decision algorithm to evaluate grid performance indexes comprehensively, find out the best allocation scheme, realize the target of grid resource allocation.Grid computing is new network computing mode, network factors is an important aspect for grid performance because of the wide-area resource. The network performance of grid including many aspects, for example: IP, network bandwith, the configuration of router, time-delay. We proposed and designed TOPSIS multiple attributes decision model to evaluate the network performance of grid node, TOPSIS defines one degree to measure some solution near ideal solution and far away negative ideal solution. According to every appraise scheme and optimum scheme approaching degree to order as the basis of evaluation superior or inferior. The scheduler evaluates grid resource performance according to TOPSIS, and uses the evaluation result in grid resource allocation decision , finally, we get the resource allocation scheme with the shortest task completed time.The reliability of grid resource is the endurance and stability when resource executes task in the process of grid resource allocation. We bring resource reliability into grid resource allocation, decrease the general cost of task execution, realize the load balancing of resource. This paper proposes Entropy Weight Coefficient Method multiple attributes comprehensive evaluation model to evaluate grid resource reliability , Entropy Weight Coefficient Method quantizes information between every choosing scheme and subjective information and builds comprehensive evaluation decision-making model. We apply Entropy Weight Coefficient Method to grid resource allocation, and use the evaluation result in grid resource allocation decision. The innovative point in this paper is: Multiple attributes comprehensive evaluation decision model is applied in grid resource allocation. We put up a wide-area and heterogeneous grid simulation experiment environment by means of GridSim simulator. We modify body entity in GridSim simulator, add TOPSIS and Entropy Weight Coefficient Method into scheduler, extend the application range of scheduling algorithm. The experiment has obtained the satisfactory result, verified the correctness and feasibility of algorithms. The results show that multiple attributes decision algorithm can evaluate comprehensively the network performance and reliability of grid. The best allocation scheme according to the evaluation results solves the resource allocation problem in the large-scale grid system.