《计算机应用》唯一官方网站 ›› 2023, Vol. 43 ›› Issue (S2): 152-156.DOI: 10.11772/j.issn.1001-9081.2023050667

• 网络与通信 • 上一篇    

基于逻辑回归的数据中心网络流量预测

万欣1(), 黄翔2, 王甫志1   

  1. 1.国能大渡河大数据服务有限公司,成都 610000
    2.国能大渡河流域水电开发有限公司,成都 610000
  • 收稿日期:2023-06-01 修回日期:2023-07-03 接受日期:2023-07-11 发布日期:2024-01-09 出版日期:2023-12-31
  • 通讯作者: 万欣
  • 作者简介:万欣(1983—),男,四川乐山人,高级工程师,主要研究方向:水电设备技术与管理、智慧企业建设
    黄翔(1980—),男,四川荥经人,高级工程师,博士,主要研究方向:水电环保管理、智慧企业建设
    王甫志(1982—),男,湖北随州人,高级工程师,硕士,主要研究方向:梯级水电站经济运行、生产运营管理、水情与电力数据分析挖掘。

Data center network traffic prediction based on logistic regression

Xin WAN1(), Xiang HUANG2, Fuzhi WANG1   

  1. 1.Guoneng Dadu River Big Data Service Company Limited,Chengdu Sichuan 610000,China
    2.Dadu River Hydropower Development Company Limited,Chengdu Sichuan 610000,China
  • Received:2023-06-01 Revised:2023-07-03 Accepted:2023-07-11 Online:2024-01-09 Published:2023-12-31
  • Contact: Xin WAN

摘要:

随着信息化发展,数据中心网络承载的通信量呈爆炸式增长,对网络带宽的需求不断增加。针对传统的网络流量预测模型精度较低和泛化能力较弱等问题,结合国能大渡河数据中心的实际情况,从时间序列预测层面,基于长期参数估计提出一种基于逻辑回归的网络流量预测方法。所提方法根据最大似然和误差相关准则得出精确解,并为数据中心实时资源分配提供指导。

关键词: 数据中心, 网络流量, 逻辑回归, 长期参数估计, 最大似然性, 误差相关准则

Abstract:

With the development of informatization, the traffic carried by data center network is exploding and the demand for network bandwidth is increasing. For the problems such as low accuracy and weak generalization ability of traditional network traffic prediction model, taking Guoneng Dadu River Data Center as an example, a network traffic prediction method was proposed based on long-term parameter estimation from a time series forecasting perspective. The proposed model can provide exact solutions based on maximum likelihood and error correlation criterion, and provide guidance for real-time resource allocation in data center.

Key words: data center, network traffic, logistic regression, long-term parameter estimation, maximum likelihood, error correlation criterion

中图分类号: