摘要
数据中心产业用户规模的不断增加和企业级应用的持续升级,导致数据中心规模不断扩张,随之而来的能耗和资源利用率问题逐渐成为研究热点。目前传统的数据中心节能策略中主要存在监控资源单一化和基于积分方式的能耗模型建立所导致的节能效率不高的问题。针对以上问题,提出一种基于神经网络的数据中心能耗模型研究,该架构包括实时自动化监控模块、能耗模块和基于神经网络算法的资源分配模块,最终利用神经网络实现数据中心资源的高效分配。并在实验室进行仿真测试,得到能耗节省率在30%以上。
The increasing scale of users in the data center industry and the continuous upgrading of enterprise-level applications have led to the continuous expansion of the data center scale, and the consequent energy consumption and resource utilization problems have gradually become research hotspots. At present, the traditional data center energy-saving strategy mainly has the problem of low energy-saving efficiency caused by the simplification of monitoring resources and the establishment of energy-based models based on the integral method. In view of the above problems, this paper proposes a data center energy consumption model based on neural network algorithm. The architecture includes real-time automatic monitoring module, energy consumption module and resource allocation module based on neural network algorithm. Finally, the neural network algorithm was used to realize the efficient allocation of data center resources. The simulation test was carried out in the laboratory, and the energy saving rate was over 30%.
作者
智伟威
周新星
ZHI Wei-wei;ZHOU Xin-xing(Aerospace Communication Center,Beijing 100830,China)
出处
《计算机仿真》
北大核心
2020年第10期273-277,共5页
Computer Simulation
基金
基于立体图像智能分割技术应用的算法研究(2018XJY01)。
关键词
神经网络
数据中心
能耗
Neural network
Data center
Energy consumption
