AMOPE: Performance Analysis of OpenFlow Systems in Software-Defined Networks

In this paper, we address the problem of defining probabilistic bounds of packet flow through an OpenFlow switch in software-defined networks (SDNs). The problem is challenging, as OpenFlow is one of the popular southbound application programming interfaces, which enables controller-switch interaction. The related existing literature addresses the different aspects of OpenFlow and SDN-controller interactions. However, there is a need to analyze the performance of the OpenFlow switch, in order to determine the bounds of the performance measures. In this paper, we propose Markov chain-based analytical model, named AMOPE, for analyzing packet flow through an OpenFlow switch, while defining the probabilistic bounds on performance analysis. Additionally, in AMOPE, we propose a state diagram based on the OpenFlow specification version 1.5.0, and calculate the theoretical probabilities of a packet to be in different states of the OpenFlow switch. Furthermore, AMOPE defines the theoretical bounds of OpenFlow performance measures such as the output action, packet drop, and send to the controller probabilities. Simulation-based analysis exhibits that approximately ${\text{60}}\%$ of the processed packets are sent to output action, ${\text{31}}\%$ of the processed packets are sent to the controller, and the remaining processed packets are dropped in an OpenFlow switch.