An Analytical Model of Data Plane Performance Subject to Prioritized Service

摘要

Software Defined Networking (SDN) is emerging as a new paradigm in which the control plane is decoupled from the data plane. SDN is an enabler of network control to become directly programmable. The underlying infrastructure can be abstracted from applications and network services. Recently, SDN has rapidly expanded and attracted many research efforts. In the open literature, analytical tools have been proposed to study the performance of OpenFlow networks to meet the demands of the rapid development of commercial deployment. Jackson model is frequently employed to characterize the features of data plane forwarding. However, network traffic frequently exhibits self-similar characteristic that has been proven repeatedly in traditional networks. Analytical models without taking the self-similar nature into account may lead to unexpected results. This paper establishes a simulation and collects traffic based on Mininet. The Hurst parameter estimation suggests that self-similar nature consists in SDN traffic. To this end, a prioritized service model subject to self-similar traffic flows is developed. A decoupling approach is applied to isolate the interacted traffic flows. Hence, the performance can be obtained by examining a single serve queueing system. Extensive comparisons between the analytical and simulation results suggest the accuracy and feasibility of our work.