Research On Job Scheduling And Resource Allocation In Grid Computing Market Environment

Grid is a new type of information technology infrastructure based on the Internet, which goal is to seamlessly integrate wide area resources to solve problems and achieve the comprehensive sharing of computing resources, storage resources, communication resources, software resources, information resources and knowledge resources. The traditional job scheduling and resousrce allocation method is ineffective for the distributed, heterogeneous, dynamic, and autonomous resources in the Grid.Firstly, the dissertation introduces the concept, evaluation and classification of Grid Computing. After analyzing the main problems in Grid job scheduling, Chapter 2 recalls theory of scheduling and clarifies the new features in Grid scheduling, i.e., large-scale heterogeneous machine, dynamic and unreliable environment and user-centric objective. In particular, the simulator OppSim is introduced.Chapter3 proposes a grouped-scheduling strategy for coarse-grained Grid application. The similar jobs are grouped and dispatched to appropriate computing resources. Compared with the fine-grained form in which each job is processed individually and then sent back to the user, this strategy can effectively decrease the overhead time and increase the processing efficiency. We use the GridSim Toolkit to testify its advantage with simulation experiments.In Chapter 4, we analyze three essential objectives for a commercial computing service based on the Grid economy environment: meet SLA; maintain reliability and earn profit. Then we developed two evaluation methods that are separate and integrated risk analysis to analyze the effectiveness of resource management policies in achieving the required objectives. We also designed an admission control strategy to maximize the providers' benefit and used the GridSim Toolkit to testify its availability with simulation experiments. From the results, we can see this strategy increase the resource providers' benefit effectively.According to the characteristics of Grid environment, combined with the existing resource searching mechanism, Chapter 5 proposes a layered framework of Grid resource searching mechanism. The Grid resources are organized in two layers, Grid Virtual Community Layer and Resource Layer. In accordance with the resource attribute, the Grid is divided logically. Based on the layered framework, according to Small-world theory, it proposes a resource discovery strategy based on the recommendation of information sharing mechanism. That is, through a third party node, two nodes which at first have no relationship can link together. They have similar demanding and can share each other's information of resource pool. The simulations show that this strategy is more efficient than Globus MDS.In Chapter 6, the LSVN (Limitation of the Scope of Visited Nodes) problem is defined, which is caused by the scope limitation of visited nodes in Grid environment. It is found that at least one third nodes in Grid are not visited. Problems caused by LSVN such as network bottleneck and service delay are studied. Three strategies based on historical information are proposed to solve the LSVN problem and promote Grid utilization. Simulation shows that all the strategies efficiently expand the scope of visited nodes and improve the Grid utilization by two or three times.By analyzing the characteristic of Grid such as large-scale, heterogeneity,unreliability and user-centric, the dissertation gives deeply analysis and beneficial practice on the typical Grid environment, and shows the brand new view and fine perspective for the future of the Grid.