Research On QoS-constrained Task Scheduling Algorithm For Grid Workflow

Grid workflow is the combination of the grid and workflow management which provides a highly promising solution for the design of the next generation of the distributed and collaborative workflow system. Grid workflow scheduling, as one of the key issues in the Grid workflow management, can have significant impact on the performance of the system. On the one hand, the Grid is an opened, dynamic network environment. Users can submit the task to the Grid everywhere. On the other hand, the Grid resources, users and the management policy can change dynamically. In these situations, the QoS-constrained task scheduling algorithm for Grid workflow is faced with the tremendous challenge.In this dissertation, the basic concepts of Grid and Grid workflow are introduced, and the scheduling problems in typical Grid workflow management systems and several typical Grid workflow scheduling algorithms are analyzed. Furthermore, we present the scalable QoS (Quality of Service) parameters structure model of the grid service, and define the utility updating functions of its multidimensional QoS parameters. Finally, we propose the QoS estimation algorithm for Grid workflow. Then we improve the MOPSO (Multiple Objective Particle Swarm Optimization), and model the workflow as the AGWL (Abstract Grid Workflow Language) with a rich set of constructions. On the basis, we propose the QoS-constrained task scheduling algorithm for Grid workflow and the best compromise option with Fuzzy set theory to solve workflow scheduling problem in grid environment. At the end of this dissertation, the experiment and theoretical analysis show the feasibility and effectiveness of the algorithm with a typical Grid workflow instance.