A review of the role of latency in multi-controller placement in software-defined-wide area networks

Abstract

Software-Defined Networking (SDN) is a computer network technology that physically separates the traditional switch architecture’s control and data planes into two separate systems enabling the configuration of SDN compliant devices to become centralised and independent via the SDN controller. Essentially, the separation simplifies an administrator’s task of managing a computer network and is cost-effective due to purchasing inexpensive switches with low processing power. Networks are then dynamically scalable as they can be partitioned into multiple data planes and controllers with policies that can be synchronised across the board. However, researchers have discovered that, in SDN, the location of these controllers impacts a network’s performance capabilities and furthermore, the control plane provides fault management and performance in SDN. Therefore, knowing where to deploy and how many controllers are needed is fundamental. This paper surveys the recent literature on this ongoing issue known as the Controller Placement Problem (CPP), its application to the Wide Area Network (WAN), and the role latency plays in it.