对称性的起因:非业务性网络流量新特性的挖掘与探索

Cause of Symmetry for Traffic Volume: Mining and Exploring New Characteristics of Traffic in Non-Business Networks

针对当前非业务性网络前后向流量呈现对称性的问题,提出一种基于双向网络流模型的分析方法。与传统的单向网络流模型相比,双向网络流模型以主机为粒度实现流量双向聚合,利用网络流在协议、端口分布和流对称性方面的特征指标刻画主机间的交互行为,从而发现流量对称性产生的根源。基于西安交通大学校园网流量的实验结果表明,1%的主机承载了90%以上的前向UDP流量,是造成流量对称性的根本原因。相比于传统的网络流模型,文中提出的模型和分析方法能快速定位造成流量对称性的主机,更适用于非业务性网络流量的控制与管理。

Measurement methods based on bidirectional flow model are proposed to investigate the symmetric characteristic of the nonbusiness network traffics. The features of protocol, port distribution and flow size are extracted from hostlevel aggregated traffic to characterize users’ interactive behavior and to find the origins of traffic symmetry. The analysis results using actual traffic traces collected from the campus network of Xi’an Jiaotong University show that 1% of the total hosts bear more than 90% of the forward UDP traffic, which results in the symmetry of total traffic. Compared with the traditional flow model, the proposed methods can quickly locate the hosts that are responsible for traffic symmetry, and are more suitable for traffic control and management in nonbusiness networks.