面向多维度QoS的云资源在线调度优化研究  

Optimization of cloud resource online scheduling on multidimensional QoS

作　　者：施进发[1] 焦合军[2,3] 

机构地区：[1]郑州航空工业管理学院管理科学与工程学院,河南郑州450015 [2]河南工程学院计算机科学与工程系,河南郑州451191 [3]西安理工大学计算机科学与工程学院,陕西西安710048

出　　处：《计算机工程与设计》2013年第12期4299-4303,共5页Computer Engineering and Design

基　　金：国家自然科学基金项目(70971120);航空科学基金项目(No.2012ZG55023);河南省创新型科技人才队伍建设工程基金项目(114200510003);郑州市创新型科技人才队伍建设工程基金项目(10LJRC183)

摘　　要：针对云计算环境下资源调度未能全面考虑QoS需求的问题,提出了一种多维QoS约束的调度方法。采用公式化描述和性能分析对多维QoS进行建模,在优化方法中,引入系统指标,通过偏好因子改变权重,结合主元和客元的合成以及公平性评判,实现了对多目标优化问题的非劣最优解的锁定。实验结果表明,该方法充分考虑了用户多维QoS需求,在系统开销相当的情况下,负载均衡程度明显提高,且平均提高了17%,资源分配能更好的符合用户满意度要求。Aiming at the problem of incomplete consideration of QoS requirements in online cloud scheduling, a scheduling approach with multidimensional QoS constraints is put forward. The formal definition and the performance analysis is employed in the modeling of multidimensional QoS. In order to fix Pareto optimal solution, system index is introduced by changing the weights in preference factor based on the synthesis of principal component and factor as well as fairness judgement. The experiment result shows that the approach gives sufficient consideration of multidimensional user QoS requirements. With the same system costs, the extent of load balance is significantly improved by an average of 17%. It can help allocate resources to better serve user satisfaction.

关 键 词：云计算 在线调度 多维QOS 偏好因子 负载均衡 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]