The Research Of Grid Resource Scheduling Algorithm Based On Utility Function

Advances in networking technology and computational infrastructure make it possible toconstruct large-scale high-performance distributed computing environments, in order to meetthe increasing requirement on computational power. Metacomputing is wide area high-performance technology that share computation resource,memory resource,data resource andother resource for implementing high-performance computing and achieve importantapplication.So metacomputing is named of grid computing technology.Because of the characters of distributed and shared, self-resembling, dynamic and diverse,autonomous and manifold management of grid, resource management and scheduling undergrid environments is one of the key research fields of grid systems.Based on the previous research, the thesis proposes the idea that using economic principlesin the management of resources in the grid. The first important contribution of microeconomicsto resource management in the Grid systems is a methodology for limiting the complexity ofsolving resource allocation problems. The second is a set of mathematical models that can yieldnew insights into resource sharing problems.Firstly, the thesis concludes the research on grid and discusses the rationality of usingeconomic method and agent technology to manager grid resource. Secondly, the thesis proposesa Service Broker-oriented Architecture for Grid Resource Management that supports highefficient and quality of services based resource management and scheduling. Our modelmatches well with real-life administrative structure of a practical Grid computing platformwhich is open and owned by a large number of autonomous management units. Thirdly, nowmost Grid job management systems focus on the optimization of system level schedulingstrategy or algorithms from the point of view of traditional batch job management, and lack inefficient supporting for dynamic changing of service Grid application.The Thesis designsMultiple QoS based resource scheduling algorithm (MQRS) and evaluates the performance ofalgorithms through a series of simulations. Comparing with other algorithms, the performanceof the MQRS algorithms shows that the economy achieves better performance than otheralgorithms and matches well with the service on demand in Grid.