Analytical model for Software Defined Network delay estimation

Traditional IP networks are becoming more complex to satisfy the demands of modern applications and it draws communication providers to exploit a wide range of specialized hardware. Commissioning of a new network service implies additional room for new equipment, power sources, cooling systems, etc. Besides, there is need for increase of qualified stuff. This results in the escalation of capital and operational expenses that eventually exceeds the revenue growth rates for which these expenses are initially intended. To overcome such a deadlock, it is no more sufficient to introduce new protocols for network enhancement, it is necessary to reconsider the network architecture. Software-Defined Networking (SDN) is a new network paradigm that allows managing network services by separating data and control planes. Herein network planning and traffic management are performed by software what makes installation of new network services, such as routing, resource management and security more simple and agile. Interaction between the two planes can be provided by means of OpenFlow protocol that allows manipulating data plane and collecting various statistics used to form global state of a network. In this paper we present an overview of SDN and OpenFlow, their key features and main concepts. Next, we present a model of SDN-network in the form of a queuing network and propose the approach to mean delay time estimation. © 2016 IEEE.