Optimum placement of SDN controllers in African backbones: SANREN and ZAMREN as a case study

Abstract

Software Defined Networking (SDN) has emerged as a potential solution to the ICT inequality challenge in emerging markets. This technology promises to revolutionize the telecommunications industry by introducing decoupled architectures to facilitate network management and configuration. A consensus was reached that a huge portion of OpEx comes from the cost associated with the management and configuration of tightly coupled legacy networks. This has contributed to operators’ reluctance to extend broadband coverage to the poor rural areas and sparsely populated areas due to the potential low profit margins. SDN opens unprecedented opportunities such as non-discriminatory infrastructure sharing, hardware commoditization (through the use of cheap commodity hardware), and business agility. This is likely to encourage operators to cover rural areas with low or no network footprint. At the heart of SDN is a controller with a global view of the current network status. It is critical that this controller is placed in a manner that optimizes network performance. This design choice is commonly known as the controller placement problem (CPP). This paper proposes an algorithm for placement of the controller that optimizes network performance, particularly propagation latency. The algorithms are tested on two African backbones, namely SANREN and ZANREN.