Grid Resource Management Based On Mobile Agent

As a sharing computer resource in large scale area solution, grid is one of the hottest research fields in recent years. Because of the wide area distribution of computer resource, the instability of the network environment and restrict of Internet bandwidth and communication delay, the grid resource discovery in wide area distribution affects the function and the performance of the grid. The mobile agent technology combines with the agent technology and the distribution technology is of most concern in the research of distribution system because of its mobility, autonomy, reactivity and conceitedness.In this thesis, the resource management technology and the research field of grid technology is introduced, including the description method of grid resource, the grid resource management mechanism, the key technology of mobile agent and the application of mobile agent in grid resource management. Based on the conclusion of current research, a novel grid resource management model based on the mobile agent technology is presented in this thesis. The novel resource management method reduces the operation width by dividing the grid environment into clusters when managing the grid resource. Based on the grid clustering, the mobile agents will choose the idle route in the network to collect, update and caching the resource information because of the autonomy of mobile agent. According to the requirement of the grid resource management system, a vector based resource description method is presented in this thesis; it's in charge of the implementation of matching and filtering the recorded information of service resource. The cached resource records in different clusters work as inspiring factor of resource discovery algorithm when the discovery operation of the grid resource is run proactively. So that the speed of the gird resource discovery can be improved, and the accuracy of grid resource discovery can be improved by using the vector based resource filtering algorithm, so that the usage performance of the grid can be increased.In the end of the thesis, experiments are run on the Aglets platform to prove the availability of the grid resource registering, discovery and filtering operations in single cluster. The Gridsim simulation software is employed to test the performance of the global grid resource discovery algorithm. The availability of the key algorithms is proved by analyzing the result of the simulations and comparing to the existing research results. The model presented in this thesis is available by the analysis of both simulations'results.