Performance Analysis of Ryu-POX Controller in Different Tree-Based SDN Topologies

Abstract

Next generation networking architecture is required to be reliable, scalable, flexible, secure and has other advanced features. Traditional TCPIP networks are complex and cannot meet the requirements for high-quality network services. Software Defined Network (SDN) is an important technology that enables a completely new approach in how we develop and manage networks. SDN divides the data plane and control plane and promotes logical centralization of network control so that the controller can schedule the data in the network effectively through OpenFlow protocol. In this paper, we simulate the two SDN controllers of Ryu and POX, and compare their latency and throughput performance under Simple-Tree-Based (STB) and Fat-Tree-Based (FTB) network topologies. An SDN networking model has been designed using a Mininet emulator, and the code for custom STB/FTB topology is executed in Python script. Simulation outcomes indicate that in latency mode Ryu controller exhibited better results than POX controller, making it more suitable for small-scale SDN deployments. From the throughput simulation, POX controller displayed better results than Ryu, showing that it is able to respond to requests more promptly under complex FTB traffic loads, but with more hardware resources utilization.

Keywords

• network topology
• next generation networking
• tree data structures
• software defined networking
• soft switching