异构云中综合时间能耗成本的任务调度算法

摘要

针对异构云环境下不合理的任务-资源映射而导致依赖任务在运行过程中产生高能耗的问题,提出一种综合时间能耗成本的任务调度算法( Time and Energy Consumption Cost Scheduling,TECCS)。根据任务图逐层进行任务调度,面对同一层任务调度顺序只单独基于时间因素考虑而过于单一的问题,引入通信因子和计算因子,综合时间与能耗成本决定同一层任务的调度顺序；分析任务之间的依赖关系,自上而下,为任务分配计算节点,使得整个任务在期望完成时间条件下节省更多能耗。从性能和能耗方面将TECCS与TUGS( Time Unify Greed Scheduling)、CATS( Communication-Aware Task Scheduling)、EETDS ( Energy Efficient Task Duplication Scheduling)进行比较,结果表明TECCS在满足任务期望完成时间条件下能耗最少。%Facing the problem of high energy consumption produced by unreasonable task-resource mapping inheterogeneous cloud,pro-pose a task scheduling algorithm based on time and energy consumption ( TECCS) . Dividing the hierarchical of the tasks to determine the order of tasks,and facing the problem of the scheduling sequence of the same layer tasks based solely on time factors into consideration and too one-dimensional,introduce communication factor and computation factor,integrating time and energy consumption cost to deter-mine the order of the same level tasks. Analysis of dependencies between tasks,based on hierarchical top-down to task scheduling,make the whole task completion time under the expected conditions to save more energy. Comparison on performance and energy consumption is conducted for TECCS with TUGS( Time Unify Greed Scheduling) ,CATS( Communication-Aware Task Scheduling) ,EETDS ( Ener-gy Efficient Task Duplication Scheduling) . Experimental results show that TECCS saves much energy under the condition of task desired time.