一种满足能效的云任务调度算法

A cloud tasks scheduling algorithm satisfying energy-efficiency

提出一种高能效的云任务调度算法。首先,计算任务优先级,根据优先级得到任务调度次序,并以最早完成时间或最小计算时间为原则,依次为任务选择调度资源形成初始调度方案。然后,计算初始调度方案下的资源能效值,从资源集合中移除能效值最低的资源,并关闭该资源。最后,更新可用资源集,进行任务重调度,即重新以最早完成时间或最小计算时间为原则为任务选择最优调度资源,完成任务调度。算例测试结果表明,算法在资源同质和异构条件下均可在不增加任务调度跨度的同时降低任务执行的总体能耗,节省资源的能源开销。

A high energy-efficiency cloud tasks scheduling algorithm is presented.First,the task priority is computed,and the task scheduling order is obtained according to the task priority.Based on earliest finish time or minimum computing time,our algorithm selects the resource for the scheduled task in turn and generates an initial scheduling scheme.Then,the resource energy-efficiency value of the initial scheduling scheme is computed and the resource with the lowest energy-efficiency value is removed from the resource set.And the resource is turned off.Finally,the available resource set is updated and the tasks are re-scheduled,i.e.,based on earliest finish time or minimum computing time once again,the algorithm selects the optimal scheduling resource and finishes the tasks scheduling.The example test results show that,under the condition of homogeneous resource and heterogeneous resource,our algorithm can reduce the total energy consumption of tasks execution and save the energy of resource without increasing the tasks scheduling makespan.