Research On Adaptive Grid Workflow Scheduling Based On Deadline Mechanism

Now the technology of grid computing and cloud computing is developing rapidly, anumber of complex applications related to the grid can be solved thought the gridworkflow platform. Resource sharing technology can be comprehensively supported bygrid technology, because of the grid resource dynamic and heterogeneous characteristics,it may cause many complex problems which can be better solved by grid workflowtechnology. However the needs of user about task deadline have become to be a majorchallenge problem in grid workflow scheduling.Traditional security authentication method in the grid environment can no longermeet the problems brought about by the dynamic and heterogeneity characters of gridresources. In this paper, I use random model to describe the dynamic nature and workloadof grid resources during the workflow scheduling process, and research the differentinfluence of workflow scheduling algorithm under different workload. In this paper aworkflow scheduling algorithm named Deadline-based scheduling on domain andadaptive dynamic grid workflow is proposed, in which we calculate the probability ofevery gird resource node about finishing workflow scheduling tasks, as well as calculatingthe critical path in order to develop a highly efficient and reliable resource schedulingmodel.The traditional workflow modeling includes DAG model and Petri net-based model.In order to express the characteristic of parallel processing tasks of the grid, we use theDAG model to describe the workflow scheduling of grid. Although the DAG model justcontains parallel and order workflow scheduling ways, it can describe most of theworkflow scheduling tasks. In this paper, we add LogP communication model to DAGmodel in order to calculate the priority and critical path of each task.Because of the differences in the ability of grid resources to handle the tasks, we introduce the concept of success rate under the deadline in this paper, as well as giving theway of calculating this rate. So we can dynamically partition the needs of time which theuser required and to choose the proper gird resources for task scheduling to ensure theefficient operation of the task.According to the simulation results, the algorithm of Deadline-based scheduling ondomain and adaptive dynamic grid workflow is better able to meet the user demand oftime constraints of the task scheduling and to handle the task of grid workflow schedulingin dynamic resource workload environment and dynamic heterogeneous grid resourceenvironment.