组合能耗优化与QoS的异构集群负载分配方法

摘要

计算能效比的异构性使得集群系统在进行任务分配时,权衡服务质量与能耗优化变得更复杂.为此,针对异构集群的任务分配问题,建立能耗优化任务分配模型,然后提出一个任务分配算法(LAOE算法),该算法在不违背任务最大平均响应时间的前提下,将任务优先分配到能效比最高的服务器.实验结果表明:LAOE算法能显著节省能耗,并能保持相对稳定的平均响应时间,具有更广泛的异构环境适应性及计算复杂度开销小等优点.%The heterogeneity of computing efficiency ratio of the cluster system made the balance of service quality and energy consumption more complexin assigning task allocation. For the task allocation of heterogeneous clusters, a task al?location model based on energy optimization was represented, and a task allocation algorithm (LAOE algorithm) was pro?posed. Using this algorithm, without violating the maximum mean response time of tasks, the task was preferentially as?signed to the server with highest energy efficiency ratio. The experimental results show that LAOE algorithm can signifi?cantly save energy consumption, and make the mean response time relatively stable. This algorithm has more extensive heterogeneous environment adaptability and lower computing complexity overhead.