The Research On Scheduling Strategy Of Scientific Workflow In Cloud Computing Environment

Cloud computing technology can reduce operating costs and ensure more stable and efficient operation of infrastructure.There are many tasks in scientific research that are computationally intensive and need to deal with a large number of dependent data.Workflow technology can automate the execution of complex scientific tasks,which is called scientific workflow.Scientific workflow needs to calculate and store massive data,so it can only be deployed in distributed environment.Distributed cluster system or grid system has high cost,while cloud computing has the characteristics of low cost,high performance,unlimited computing and storage resources,so scientific workflow is generally executed in the cloud environment.Combining the new characteristics of cloud computing to develop efficient workflow scheduling methods has become an important research direction.This thesis mainly studies the scheduling of scientific workflow in cloud computing environment.The main research contents of this thesis include the following aspects:(1)Modeling the scheduling and execution of scientific workflow tasks,discussing the challenges and improvement directions that existing workflow scheduling algorithms and execution frameworks may encounter.(2)The scheduling problem of scientific workflow is studied and two scheduling methods are proposed.The first scheduling method is workflow scheduling method with budget and deadline constraints,which can schedule scientific workflow with both deadline and budget constraints.Experiments show that the algorithm has good performance.The second scheduling method is fault-tolerant workflow scheduling method based on deadline.This algorithm takes into account the virtual machine failure in scheduling,and gives priority to the sub-tasks on the critical path,which shortens the overall completion time of the whole workflow application.(3)Based on OpenStack and Pegasus workflow scheduling system,the scheduling method is developed into workflow scheduling module,which verifies the feasibility of the algorithm.