海量数据存储中云服务器性能加速方法仿真  

Simulation of Cloud Server Performance Acceleration Method in Massive Data Storage

作　　者：危华明 廖剑平 WEI Hua-ming;LIAO Jian-ping(Nanning University,College of Information Engineering,Nanning Guangxi 530200,China;Nanning Normal University,School of Computer and Information Engineering,Guangxi Nanning 530299,China)

机构地区：[1]南宁学院信息工程学院,广西南宁530200 [2]南宁师范大学计算机与信息工程学院,广西南宁530299

出　　处：《计算机仿真》2023年第5期515-519,共5页Computer Simulation

基　　金：广西高校中青年教师科研基础能力提升项目(2020KY64017)

摘　　要：用户将海量数据存储于云服务器以节省本地存储空间,但是由于云数据的种类复杂、数据量庞大,严重影响了云服务器整体存储的速度。基于此,提出考虑海量数据存储的云服务器性能加速方法。基于小波分析理论,对海量存储数据小波包分解处理,消除噪声数据。构建数据存储结构模型,提取关键信息特征。应用卷积神经网络,设计CNN云服务器加速存储模型,实现云服务器性能加速。实验结果表明,上述方法的吞吐量指标在95%-100%之间,CPU利用率在85%以上,可准确预测云服务器负载,实验所得数据验证了所提方法的可靠性高、应用有效性更强。At present,some users store massive data on cloud servers to save local storage space.Due to the complex types and massive cloud data,the storage speed of the cloud server is seriously affected.Therefore,a method of cloud server performance acceleration considering massive data storage was proposed.Based on the theory of wavelet analysis,the wavelet packet of massive storage data was decomposed to eliminate noise data.And then,a model of data storage structure was constructed to extract key information features.Finally,a CNN accelerated storage model was designed by convolutional neural networks,thus accelerating the performance of the cloud server.Experimental results prove that the throughput index of the proposed method is between 95%-100%,and the CPU utilization is more than 85%.In addition,this method can accurately predict the load of the cloud server.The experimental data verify the high reliability and application effectiveness of the method.

关 键 词：海量数据存储 云服务器性能加速 卷积神经网络 小波包分解 特征提取 

分 类 号：TP3[自动化与计算机技术—计算机科学与技术]