满足实时云任务需求的主动响应式工作流调度模型

AN ACTIVE AND RESPONSIVE WORKFLOW SCHEDULING MODEL MEETING THE REQUIREMENT OF REAL-TIME CLOUD TASKS

为了改善云平台中多工作流调度时的执行代价和资源利用率,提出一种满足实时云任务需求的主动响应式工作流调度算法。该算法可以分别针对新工作流的到达、任务完成后的虚拟机资源提供、紧迫任务到达等做出主动的实时响应调度策略,从而充分利用虚拟机资源的空闲时槽和更大化的任务并行程度,以混合形式调度来自不同工作流的任务。在确保截止期限约束的同时,有效满足实时云任务的调度需求。大量实时工作流的仿真测试表明,与另外几种同类型算法相比,该算法不仅可以降低任务执行代价,而且可以得到更高的资源利用率,实现算法预期效果。

To improve the execution cost and the resource utilization of workflow scheduling in cloud platform,an active and responsive workflow scheduling algorithm meeting the requirement of real-time cloud tasks is designed in this paper.According to the arrival of the new workflows,the virtual machine resource provision after tasks completion and the arrival of urgent tasks,this algorithm could make an active and real-time responsive scheduling strategy.It could take full advantage of idle time slots of virtual machines resource and the greater task parallelism,and the tasks from different workflows were scheduled in a hybrid way to meet the scheduling requirement of real-time cloud tasks while guaranteeing the deadlines of workflows.The simulation tests of lots of real-time workflows show that,compared with other same types of algorithms,this algorithm can not only reduce tasks execution cost,but also obtain higher resource utilization,which obtains an expected outcome.