HawkFlow: Scheme for Scalable Hierarchically Distributed Control in Software Defined Network

Abstract

Compared with the traditional design of network architecture, Software Defined Network (SDN) can be programmed to provide more flexible, fine-grained and differentiated services because of its control centralization property. However, with a variety of network functions such as firewall and multicast are gradually added to the SDN controller, the heavy computational load on SDN control plane has made it the bottleneck of whole network architecture in large scale networks. Among all the solutions proposed in the literature, distributed control plane is very promising in solving the problem. This paper presents HawkFlow, a scheme based on hierarchically distributed control plane, to improve the efficiency and scalability of SDN control plane. HawkFlow proposes blocking island theory and network aggregation mechanism to reduce the searching space of Centralized Single Controller Routing (CSR) algorithms. Routing requests are divided into three levels according to the destination IP address, in which routing processes in local networks are designed to be CSR to reduce the average routing complexity. Experimental results show that the mechanisms discussed in this paper can greatly improve the efficiency of hierarchically distributed control plane, especially in the networks with large proportions of local network traffic such as data centers or campus networks. 1