A Low-Cost Measurement Framework in Software Defined Networks

• 作者: Qiang He Shengbao Wang
• 来源期刊: 通讯、网络与系统学国际期刊(英文)
• 学科: 工学
• 关键词: SOFTWARE Defined Network ADAPTIVE Sampling POX CONTROLLER OpenLL
• 年，卷（期）: 2017,（5）
• 页码范围: 54-66
• 页数: 13页
• 分类号: TP39

摘要

Software Defined Network (SDN) makes network management more flexible by separating control plane and data plane, centralized control and being programmable. Although, network measurement still remains in primary stage in SDN, it has become an essential research field in SDN management. In this context, this paper presents a low-cost high-accuracy measurement framework to support various network measurement tasks, such as throughput, delay and packet loss rate. In this framework, we only measure per-flow edge switches (the first and the last switches). In addition, a new adaptive sampling algorithm is proposed to significantly improve measurement accuracy and decrease network overhead. Meanwhile, we consider a low-cost topology discovery approach into our framework instead of topology discovery currently implemented by SDN controller frameworks. In order to improve the accuracy of delay, we also join a time threshold value to adjust the time delay. Furthermore, we consider and analyze the balance between measurement overhead and accuracy in several aspects. Last, we utilize POX controller to implement the proposed measurement framework. The effectiveness of our solution is demonstrated through simulations in Mininet and Matlab.