Increasing The Network Control Ability and Flexibility In Incremental Switch Deployment for Hybrid Software-Defined Networks

Abstract

Software-Defined Network is a new networking generation to increase network flexibility and performance improvement. However, budget constraints, technical limitations and the risk of architectural changes that organizations are encountered with, are barriers to the full implementation of the SDN. For these reasons, Internet Service Providers (ISPs) should replace only a limited number of conventional switches with SDN switches at one time. Put both switches together in a topology, create hybrid SDN. In this paper, we first cluster the switches into two clusters, based on a graph concept which is called closeness-centrality. Afterward, we have one cluster with low closeness centrality and another with high; then, we propose a heuristic migration scheme for deploying SDN switches in hybrid SDNs. We first allocate all existing budget to a cluster at low closeness-centrality and then we upgrade the switches according to the proposed scheme. There are four main issues in this paper: (1) improving the network control ability with a given upgrading budget constraint, (2) minimizing the upgrading cost to achieve the best network control ability, (3) increasing hybrid SDN flexibility and (4) improving network fault tolerance by deploying a link fault recovery algorithm. Also, in order to perform controller placement, we propose a simple solution to provide an acceptable delay between controller and SDN switches. We evaluate our scheme in three real topologies and one synthetic topology. The results show that our scheme can achieve better network control ability and flexibility with the help of graph theory parameters compared to other related papers according to budget constraint.