The Research Of Multi-Agents Grid Resource Management Based On QoS

Grid computing is a kind of new technology to emerge computing resources, storage resources and data resources into a virtual super-computer with powerful calculation capabilities. QoS (Quality of Service) usually refers to the network reliability and availability. Grid QoS includes not only the traditional network QoS; it also includes other QoS requirements such as CPU speed, data quality and cost of services, the deadline time. QoS ensure that applications can get the quality of service it needs. Grid Architecture needs a framework supporting Quality of Service (QoS) to manage the complex computing and data-intensive applications. Currently, the grid QoS mechanism and its implementation need further study.Agent is a kind of autonomy, social, initiative and adaptability intelligent entity. Agent-based approach will allow grid management more intelligent and convenient .Agent technology will be an effective way to solve the problem of grid. Agent technology provides technological support for the distributed system design; QoS mechanism determines the quality of the entire grid system. Agent theory and technology has major significance for the development of grid technology. This paper describes the design of a QoS-based multi-agent grid resource management model. The model has a serial of advantages: 1) the grid services are based on QoS; 2) to use agent to assess and manage QoS; 3) grid resource hierarchical management model based on Multi-Agent (MAS).This paper simulates the modified QoS Min-Min algorithm on the GridSim grid simulator. The simulation results show that the modified algorithm provides a better performance of resource scheduling than the standard Min-Min algorithm. This shows that grid with QoS has better performance than the grid without QoS services.Finally, the paper discusses some problems of the model, such as QoS Agent reliability, the application of mobile agent.