Research Of Grid Decision-making Model-base Resource Allocation Strategies On Grid Environment

Grid as a new method of calculation, because of its geographically distributed heterogeneous resources to be shared and selected, what’s more it can solve the problem of large-scale computing, such as in scientific and commercial, is getting more and more people’s attention. Grid technology is the latest technology in recent years; people regard the high performance of the resource sharing and collaborative work in the virtual network environment as the main research targets of the grid. The grid as an organization can organize the remote resources which distribute in different areas and equipment into integrated whole, and has great ability to solve the great challenges.Grid computing is different from traditional distributed computing, focusing on large-scale resource sharing and innovative applications. Resources belong to different organization in grid environment, through the organization operation, the resources spread out in different geographical location. We need to use a special mechanism to distribute the resources on different distribution site according to the problems that may occur in distributing the resources and the interrelationship between each site. Many applications which are difference in the security mechanism and the use of strategies may use the resource on the multiple sites, while different site management system may use different resources. The user’s characteristics, time, priority and execution of the goals in grid determine the utilization of the resources and cost. Therefore, the resource management and scheduling is very complicated work in the scale of the distributed environment.In this paper, we adopt the way of directory service based on the Web and LDAP to realize the comprehensive evaluation the data in decision model and the information’s organization and positioning, as well as completed the release function in each subsystem algorithm information. As the Comprehensive Evaluation decision-making Model-base Directory Management System is consisted of several distributed subsystems, we integrate information of each subsystems model to form a complete model-base system, while giving the LDAP-based data storage solutions. Directory service uses "global-local" approach to manage the resource information to improve system’s scalability and reliability. We propose a grid resource allocation strategy based on T. Cover model for evaluate the performance of grid. We use GridSim to run the simulator experiment and compare it with random algorithm. The results show that the distribution strategy achieves better performance in the completing time and cost, improves the efficiency of resource allocation, what’s more it has good adaptability in the dynamic, heterogeneous grid environment, and realizes the balancing distribution of grid resources.