基于动态阈值和占优资源的云资源调度方法研究  

Cloud resource scheduling method research based on dynamic threshold and dominant resource

作　　者：李腾耀 张月玲[1] 张凤琴[1] 严晓梅[1] 张水平[1] 

机构地区：[1]空军工程大学信息与导航学院,西安710077

出　　处：《计算机应用研究》2017年第4期1199-1203,1216,共6页Application Research of Computers

摘　　要：为了适应云资源动态变化和不确定的特性,解决静态阈值方法中调节滞后和以单资源作为度量带来的调度偏差问题,提出了以调度频率和负载率作为反馈值的动态阈值法,通过动态更新阈值确定调度的触发时机;提出了基于占优资源的调度方案,综合考虑多种异构资源对负载的影响实施调度。通过分别与静态阈值法和仅考虑单资源的调度方案进行比较,提出的调度算法分别获得了17.7%和30.7%的性能提升,这表明动态阈值法和基于占优资源的调度方案能够适应动态变化的云环境。To adapt the dynamics and uncertainties of cloud resources, this paper proposed the dynamic threshold method and dominant resource solution. Based on scheduled frequency and load ratios, dynamic threshold method was applied to choose the best scheduled time dynamically. By taking heterogeneous resource impacted on load into considerations, dominant re- source solution was applied to implement resource scheduling precisely. By the comparison with static threshold method, dy- namic threshold method gained 17.7% improvement to get over delay on scheduling time. And by the comparison with single resource solution, dominant resource solution gained 30.7% improvement to get over deviations in single resource solution. In conclusion, dynamic threshold method and dominant resource solution can adapt with dynamic cloud environments.

关 键 词：云计算 资源调度 动态阈值 占优资源 

分 类 号：TP393.07[自动化与计算机技术—计算机应用技术]