【摘要】 在软件定义网络(SDN)环境下,通过深度包检测(DPI)技术对网络流量进行分类识别。在应用层部署的DPI功能节点越多,网络流量分析的总处理时延就越小。但由于DPI节点的部署成本昂贵,因此需要优化部署方案。针对SDN中DPI节点部署问题,根据QoS处理时延,对DPI节点进行负载均衡,将该问题抽象为多商品流的线性规划问题,并设计基于成本的两阶段贪心算法加以解决。通过贪心策略为入口交换机选择低成本的DPI节点,并针对流处理过程进行负载均衡,进而通过搜素算法寻找最短路径用以降低时延。实验结果表明,该方法优化了DPI节点部署成本,提升了流量分析与检测的请求成功率。更多还原

【Abstract】 In software defined network(SDN)environment,network traffic is classified and identified by the deep packet inspection(DPI) technology. The more DPI function nodes deployed in the application layer,the smaller the total processing delay of network traffic analysis. But due to the high cost of deploying DPI nodes,it is necessary to optimize the deployment plan. In allusion of the deployment problem of DPI function nodes in SDN environment,the load balancing is performed for DPI function nodes according to QoS processing delay,and the deployment problem of DPI function nodes can be abstracted into linear programming problem of multi-commodity flow. The cost-based two-stage greedy algorithm is proposed in this paper to solve the problem. The low-cost DPI nodes can be selected for the entry switch by the greedy strategy. The load balancing is adopted for the process of flow processing to find the shortest path by means of the search algorithm,so as to decrease the delay.The experimental results show that the method has optimized the cost of deploying the DPI nodes and improved the success rate of requesting flow analysis and detection.更多还原