流量自适应的多维度包分类方法研究

发布时间：2019-02-12 14:23

【摘要】：以软件定义网络为代表的新兴网络技术发展使得包分类不再局限于传统的五元组而是面向更多维度,以开放协议OpenFlow为例,网络包分类需要针对几十个维度,并且维度数量在持续增加,这种多维度包分类功能成为了软件定义网络实际应用中的性能瓶颈.该文分析了五元组包分类算法向更多维度扩展的局限性,利用网络流的局部性原理,提出了一种流量自适应的多维度包分类方法.该方法可以根据网络流量的实时分类结果动态调整多维度匹配顺序,优先匹配当前流量所需要的字段,通过忽略通配字段达到优化查找速度的目的.同时,该方法将多维度字段分组,结合具体字段类型选择最优匹配算法.在Open vSwitch中增加所提出方法,实验结果表明,该方法相比已有的包分类算法在用户态模式下性能提高约2倍,相比从五元组包分类算法扩展的方法性能提高40%以上.
[Abstract]:With the development of new network technology represented by software definition network, packet classification is not limited to the traditional five-tuple group, but faces more dimensions. Taking open protocol OpenFlow as an example, the network packet classification needs to be aimed at dozens of dimensions. And the number of dimensions continues to increase, this multi-dimensional packet classification function has become the performance bottleneck in the application of software-defined network. This paper analyzes the limitation of the five-tuple packet classification algorithm extending to more dimensions, and proposes a new multi-dimensional packet classification method based on the local principle of network flow. This method can dynamically adjust the multi-dimension matching order according to the real-time classification results of network traffic, and first match the fields needed for the current traffic, and achieve the purpose of optimizing the search speed by neglecting the matching fields. At the same time, the multi-dimensional fields are grouped and the optimal matching algorithm is selected by combining the specific field types. The proposed method is added to Open vSwitch. The experimental results show that the performance of the proposed method is about 2 times higher than that of the existing packet classification algorithm in user mode, and the performance of the proposed method is improved by more than 40% compared with that extended from the quintuple packet classification algorithm.
【作者单位】：北京理工大学计算机学院;延安大学网络信息中心;清华大学计算机科学与技术系;
【基金】：国家自然科学基金(61272510,61432002,61522205) 陕西省教育厅科学研究项目(14JK1825)资助~~
【分类号】：TP393.0

【相似文献】