Analytical model for a fragment of SDN architecture

Автор: Mochalov V.P., Linets G.I., Palkanov I.S.

Журнал: Инфокоммуникационные технологии @ikt-psuti

Рубрика: Технологии компьютерных систем и сетей

Статья в выпуске: 2 т.18, 2020 года.

Бесплатный доступ

A fragment of a Software-Defined Networking architecture with the functionality of switches and controllers is described as the mathematical model of a multiphase queueing system with losses at each phase. In the Internet Engineering Task Force draft Address Resolution Delay in SDN (also see ITU Radiocommunication Sector M.2083-0), it was recommended to reduce network delay within a single Software-Defined Networking segment from 50 to 1 ms. This performance index depends considerably on the characteristics of the software and hardware system of Software-Defined Networks, as well as on the processes of interaction between the switches and the controller. However, the service time and hence the processing delays for packet flows are not regulated by the suppliers of switching equipment (e.g., Cisco Catalyst 3750 switches), which makes it difficult to determine the probabilistic and time characteristics of networks during the design stage and to formalize any suggestions on improving the performance of network elements. The models presented in this paper are based on the classical queueing theory and Laplace transforms. In this case, the relation between the stages of packets processing on network devices has little significance; therefore, the performance indices are obtained in terms of single-phase network parameters. The mean service time and the mean number of packets in a network are calculated as functions of the load of network devices. Also, analytical expressions for determining the mean loss ratio of network packets at each phase of processing by a switch are derived.

Еще

Протокол open flow, software-defined networking, openflow protocol, software-defined networking controller, software-defined networking switch, three-phase model of queueing systems, laplace transform

Короткий адрес: https://sciup.org/140256253

IDR: 140256253   |   DOI: 10.18469/ikt.2020.18.2.05

Статья научная