Adaptive resource allocation for software defined networking controllers

Article type: Research Article

Authors: Soursouri, Masoud | Ahmadi, Mahmood^{; *}

Affiliations: Department of Computer Engineering and Information Technology, Razi University, Kermanshah, Iran. E-mails: msd.soursouri@gmail.com, m.ahmadi@razi.ac.ir

Correspondence: [*] Corresponding author. E-mail: m.ahmadi@razi.ac.ir.

Abstract: In traditional networks, the control and data planes are integrated into a single system which makes complicated administrating and managing of these networks. Therefore, Software Defined Networking (SDN) decouples these planes using a single controller component. An SDN controller is a fundamental element that consists of applications, northbound and southbound interfaces. It also creates and manages a general view of the whole network. The increasing attempts to improve the performance of the controller have been performed, which proposed the centralized and distributed architectures for the controllers. In this paper, a service to allocate the resources to an SDN controller is proposed. Resources such as CPU and memory based on demand are allocated. In addition, to efficient use of resources when they are not needed, deallocation of resources is done. This service has been designed based on the XenServer hypervisor OpenStack platform, OpenDayLight SDN controller and their capabilities to change the resources dynamically. The results related to resource usage show that increasing the number of switches increases the number of CPUs, and memory, linearly.

Keywords: SDN controller, Cloud computing, XenServer hypervisor, adaptive resource allocation

DOI: 10.3233/JHS-170569

Journal: Journal of High Speed Networks, vol. 23, no. 3, pp. 237-253, 2017