Research On Trust Mechanism-based Resource Scheduling In Grid Environment

Grid technology is developed recently from massively distributed computing technology based on the Internet. Grid is also deemed as the computing infrastructure of the next generation of the internet in the future. It integrates geographically distributed resources into a unparalleled powerful information system. It realize the comprehensive resource sharing of the computing resources, memorizer resources and data resources etc. And the share can be controlled highly. Because the resources in grid are usually distributed, heterogeneous, and owned by different organizations, they are managed with different resource management mechanism and charging policies. The load balancing and usability of resources change simultaneously. So the management and scheduling of resources is a hard undertaking. And how to effective scheduling is a key problem for the performance of grid computing.The existing grid systems consider the authentication of resources nodes. But the authentication can't guarantee the security and reliability of the behavior. At present existing resources management in grid largely ignore the impact of trust mechanism. Most scheduling mechanisms and algorithms concern some scheduling performance parameters, they ignore the impact of the behavior trust. In fact, any one of the grid dealing entities can decrease the QoS of performance by the baleful behaviors due to the absence of trust. It also impact the efficiency of grid system.Trust relationship can divide into identity trust and behavior trust. The identity trust concerns mainly the authentication and authorization. Yet the behavior trust concerns mainly the trustiness of grid entities. This paper make a detail analyse of the trust relationships in gird and the expressions, measurement and updating of the trust relationships. We introduce the trust mechanism to the grid scheduling. Based on the traditional resources management and scheduling framework, we introduce the trust mechanism, which is used to compute and update the Trust Value of grid resource nodes. Thus trust mechanism-based resource scheduling model and policy in grid is proposed. During the scheduling, it consider not only the QoS of tasks and the performance parameters, but also the trust relationship each other. Then grid system can allocate the task to the resource nodes with high Trust Value. So it can reduce the total completion time, it also can increase the probability of the task that executing successfully. The performance of grid is improved. It also assure the well-balanced function of grid environment.This paper analyse the most classical algorithm Min-Min and point out it can lead to a bad quality of load balancing. It also doesn't concern the QoS of tasks. Based on the trust model, Min-Min algorithm is enhanced to ensure security during the scheduling, the Trust-Min-Min algorithm is proposed. It concerns not only the QoS of tasks but also the trust relationship. At last we simulate our algorithm with the aid of GridSim toolkit, we compare the algorithm to Min-Min in many aspects and the simulation result indicate that our algorithm is more reasonable and efficient. Therefore it is an efficient scheduling algorithm for scheduling in grid environment.