Visualization of elephant flows and QoS provisioning in SDN-based networks

摘要

Elephant flows in data center networks have a tendency to utilize a lot of bandwidth, leaving latency-sensitive mice flows choked behind them. This causes degradation of application performance running on the network. Therefore, a data center network should be able to not only identify elephant flows, but also offer Quality of Service (QoS) provisioning. SDN makes network operating systems to achieve greater governance of the control plane within a given network. In order to detect elephant flows, we use a framework based on sFlow sampling technology. We show an approach to guarantee QoS that is administered and defined by a centralized network controller by using SDN and specifications offered by OpenFlow. Although there have been efforts in this area over the past but in the scope of this paper, we will only focus on how Traffic Shaping (rate limiting) based classification technique can be used within an SDN network.