A Reactive Traffic Flow Estimation in Software Defined Networks
Shuangyin Ren, Gaigai Tang · 2020 5th International Conference on Computer and Communication Systems (ICCCS) · 2020
Software Defined Networks paradigm decouples network control from switches and is directly programmable. A relatively centralized control plane together with developed APIs between control and data plane allows it to retrieve a global view of network traffic. Traffic estimation algorithm, which aims to generate a realtime and precise traffic flow undergoing with acceptable overhead, is fundamental for traffic engineering and QoS guarantee. A few traffic estimation algorithms is proposed based on Software Defined Networks paradigm, such as Flowsense and Payless. We propose a reactive traffic estimation algorithm to retrieve network traffic information. This reactive traffic estimation algorithm sends two query messages after a flow table is installed, and estimates upper bound bandwidth. We run a simulation on Mininet with Ryu as controller and OVS as data plane switch to test proposed algorithm. Simulation shows that this algorithm shows a better comprehensive performance in aspect of realtime and bandwidth overhead.