云环境中期限分割下工作流调度代价优化仿真

Simulation on Workflow Scheduling Cost Optimization under Deadline Division in Clouds

为了优化云环境中期限约束的工作流调度代价优化问题,提出一种期限分割的工作流调度代价优化算法(Workflow Scheduling Cost Optimization under Deadline Distribution,WSCO-DD)。算法将工作流任务的调度过程划分为4个阶段:工作流分层、期限分割、任务选择和实例选择。工作流分层通过升秩值方法将所有工作流任务进行层次划分,提高并行执行程度;期限分割实现全局期限在不同工作流层次上的子分割;任务选择基于最早开始时间原则赋予任务优先级,得到任务调度序列;实例选择旨在选择执行代价最小的最优实例。通过科学工作流结构的仿真实验,证明WSCO-DD算法在满足期限约束的同时,在降低工作流执行代价和提高调度成功率性能上均优于其他算法。

For optimizing workflow scheduling cost under budget constraint in cloud,a workflow scheduling algorithm WSCO-DD is presented.WSCO-DD divides the workflow scheduling process into four steps:workflow leveling,deadline distribution,task selection and instance selection.Workflow leveling partitions all workflow tasks into different levels by the top-down rank,which can improve parallelism degree.Deadline distribution divides the overall deadline into the defined levels.Task selection selects the scheduled-priority task by the earliest start time principle,which can obtain the order of scheduled tasks.Instance selection selects the best instance with minimizing execution cost.Through the simulation experiments of scientific workflow,it is proved that WSCO-DD can better reduce the workflow execution cost and improve the scheduling success rate than other algorithms under contraint of meeting deadline.