Українська   Русский
DonNTU   Masters' portal

Abstract

Content

Introduction

The demand for modern networks is increasing, and at the same time, various ways of solving such problems as providing a high level of scalability with ensuring the required quality of service (QoS) are developing at a tremendous speed. At the same time, they are trying to reduce the cost of the network infrastructure and the costs of its maintenance.

Analysis of the current state of the problem

The main problem of most networks is that the set of QoS support tools was originally built by manufacturers into network devices. This, in turn, does not match the speed of the introduction of services and new applications into existing data networks.

One solution is to use the architecture of software-defined networks (Software-defined Networking, SDN). Centralization and openness of the SDN management tools allows flexible and efficient adaptation of the work of the data center to emerging business needs, which accelerates the introduction of innovations and ensures the competitiveness of companies.

For large networks of providers, the problem remains the speed of network recovery after the introduction and configuration of a new device. For example, if you start a new device, for example, the access control list (ACL) on all network devices in the network can take a long time to migrate. This is due to the distribution of control. As a result, administrators have to spend a lot of time to reconfigure the traffic processing rules on each network device. Similar problems occur if you need to reconfigure QoS mechanisms when implementing a new application, for example, video communications. Similar problems occur if you need to change the protection settings. The latter can weaken the overall security of the network.

The transition to centralized network management, implemented in software-configurable networks, allows to mitigate the aforementioned shortcomings. The main difference of SDN from data transmission networks is centralized intelligent control and network monitoring, which provides verification, control and modification of data streams. According to the SDN concept, all control logic is transferred to specialized controllers , which are able to monitor the operation of the entire network (Figure 1).

SDN architecture components

Figure 1. SDN architecture components

Goal and tasks of the research

The purpose of the work is to develop recommendations for the implementation of SDN-architecture in the existing network. To achieve this goal, it is necessary to analyze the possibility of transition to such a structure.

During the research it is necessary to solve the following tasks:
• analyze the problems of implementing SDN-solutions in the network of the Internet provider;
• develop recommendations for the evolutionary implementation of SDN-solutions in the transport network;
• Develop a methodology for estimating the parameters of economic efficiency of using SDN-solutions by the Internet provider;
• to offer means for efficient management of multiservice traffic of the SDN transport network;
• Develop a simulation model that allows you to evaluate the efficiency and effectiveness of the proposed solutions.

Analysis of the desirability of implementing SDN solutions in the network provider

The program-configurable network (SDN from English Software-defined Networking, also a program-defined network [1-5]) is a data network in which the level of network management is realized in a programmed and spatially separated manner with the transmission device. This architecture was created, due to the need to separate the functions of traffic transmission and management functions. This is connected, first of all, with the "boundedness" of the function of traffic management algorithms in existing network devices. Those. suppliers of network equipment in fact provide a certain set of such functions. In addition to this, traffic transmission processes, as well as traffic management processes are implemented in a single device (chip). Thus. it is possible to forward packets from one port to another based on rules defined by the device software (packet analysis, changes in the service information contained in them, etc.).

The SDN concept allows you to configure the network as a single system independent of the equipment manufacturer. What is achieved by creating a unified interface between the management layer and the level of data transmission. The elements of SDN-architecture are: SDN applications, SDN controller [6-10], control agents, functions that are incorporated into OpenFlow switchers [11-15], FlowVisor [16,17,27] or interface, responsible for transmission of control information, management and administration components.

The basis of SDN management is the OpenFlow protocol, which tracks changes at the data transfer level and places them in the redirection table, and also modifies and forwards control information between the controller and the switches. The OpenFlow protocol controls the exchange of messages about changes in redirection tables, while maintaining a standard set of parameters.

Openflow protocol

Openflow protocol (7 frames, infinity loops, 138 kb)

Development of recommendations for the modernization of the network

One of the main factors, both in the location of the controller, and in the choice of their number for a large Internet provider, are the delays in the controller's responses [18-20] to requests of network elements. It is very important to have a fast dynamic response to events. To solve this problem, knowing the location of the network nodes and the distance between nodes, a technique is proposed for determining the location of the controller.

Suppose that there is a graph G (V, E), where V is the number of nodes, E is the number of connections. Denote by d (r, s) the shortest path from the node r V to s V, we obtain:

Where Lср (R) is the average delay when the controller is located in r V. Comparing the average delays for each possible r, we find the optimal position of the controller [4]. If the optimal average delay obtained is not satisfactory, or because of a very large number of network elements, the operational capabilities of the server are not enough, it would be logical to use several controllers, having broken our G (V, E) network into several G1 (V1, E1) .. Gn (Vn, En) by the principle of topological proximity of nodes of the network, and apply to each subset the proposed technique. Of course, in this case, the controllers must be connected with each other. Their interaction should be based on the fact that each controller must communicate with its neighbors through a dedicated communication channel. This is achieved by installing the application on the controller, for which the controller-controller interaction turns into a version of the controller-switch.

At this stage of writing the master's work, a number of recommendations were made to introduce SDN-architecture into the functioning network of the Internet provider. In the course of further research, a new SDN-structure will be modeled using the example of the Internet provider Orion in Snejnoe and an economic efficiency assessment will be performed.

It is assumed that after the final stage SDN-architecture will have an increased capacity (stable operation is ensured even with 100% load of channels), the possibility of rebuilding the physical network without interrupting maintenance in virtual networks, an increased degree of security [21-26] due to the complete isolation of virtual networks to each other, increase network reliability due to self-recovery and automatic redistribution of traffic flows in accordance with the rules.

Thus, the transition to the SDN-architecture will improve the quality of both traffic management and the network as a whole. On the other hand, this decision will allow to reduce the price of modernization of the existing network of the provider Orion Snejnoe, and, consequently, to improve the quality of services at their constant cost.

References

  1. SDN Architecture Overview [Электронный ресурс]. – Режим доступа: www.opennetworking.org.
  2. Software defined networking [Электронный ресурс]. – Режим доступа: inau.ua.
  3. Центры обработки данных. SDN: «третий элемент» для будущих дата-центров [Электронный ресурс]. – Режим доступа: www.sib.com.ua.
  4. Introduction to Software Defined Networking (SDN)[Электронный ресурс]. – Режим доступа: www.cse.wustl.edu.
  5. Open Networking:Dell’s Point of View on SDN [Электронный ресурс]. – Режим доступа: i.dell.com.
  6. What's Software Defined Networking? [Электронный ресурс]. – Режим доступа: twiki.di.uniroma1.it.
  7. SDN-NFV Reference Architecture [Электронный ресурс]. – Режим доступа: innovation.verizon.com.
  8. Software Defined Networking (SDN) - Open Flow [Электронный ресурс]. – Режим доступа: www.ics.uci.edu.
  9. Программируемый Интернет [Электронный ресурс]. – Режим доступа: www.ripn.net.
  10. Software Defined Networking (SDN) and OpenStack [Электронный ресурс]. – Режим доступа: f5.com.
  11. Oracle SDN Performance Acceleration with Software-Defined Networking [Электронный ресурс]. – Режим доступа: www.oracle.com.
  12. SDN for Service Providers [Электронный ресурс]. – Режим доступа: www.cisco.com.
  13. Software Defined Networking Concepts [Электронный ресурс]. – Режим доступа: homepages.inf.ed.ac.uk.
  14. Software Defined Networking [Электронный ресурс]. – Режим доступа: www.comp.nus.edu.sg.
  15. SDN Software Defined Networks [Электронный ресурс]. – Режим доступа: www.control.lth.se.
  16. О.Ю. Евсеева, Е.Н. Иьяшенко, Е.Б. Ткачева Математическая модель и метод комплексного управления ресурсами транспортной программно-конфигурируемой сети // Электронное научное специализированное издание – журнал «Проблемы телекоммуникаций» №1(18) 2016 .
  17. Красотин А.А., Алексеев И.В. Программно-конфигурируемые сети как этап эволюции сетевых технологий // Модель и анализ информ. систем. №4 2013.
  18. Ю. Ю. Коляденко, Е. Э. Белоусова ПРОГРАММНО-КОНФИГУРИРУЕМЫЕ СЕТИ НА БАЗЕ ПРОТОКОЛА OPENFLOW И ИХ ХАРАКТЕРИСТИКИ // Scientific Journal «ScienceRise» №3/2(20)2016.
  19. Н.Ф. Бахарева, Ю. А. Ушаков, М. В. Ушакова, А.Е. Шухман ОСНОВЫ ПРОГРАММНО-КОНФИГУРИРУЕМЫХ СЕТЕЙ // Учебное пособие.
  20. Боклагов В.С., Лозинская В.Н. ИССЛЕДОВАНИЕ УСЛОВИЙ ПЕРЕХОДА НА SDN-АРХИТЕКТУРУ СЕТИ ПРОВАЙДЕРА «ОРИОН» Г. СНЕЖНОЕ // Автоматизация технологических объектов и процессов. Поиск молодых : сборник научных трудов ХVII научно-технической конференции аспирантов и студентов в г. Донецке 24-25 мая 2017 г. - Донецк : ДОННТУ, 2017. – 409 с..
  21. Cisco Visual Networking Index: Global Mobile Data Traffic Forecast Update, 2013–2018: Cisco® February 5, 2014 [Електронний ресурс].
  22. The Road to SDN: An Intellectual History of Programmable Networks: Nick Feamster, Jennifer Rexford, Ellen Zegura [Електронний ресурс].
  23. OpenVirteX – A Network Hypervisor : Open Networking Lab [Електронний ресурс].
  24. The Controller Placement Problem: Brandon Heller, Rob Sherwood, Nick McKeown [Електронний ресурс].
  25. OpenFlow Switch Specication: Open Networking Foundation June 25, 2012 [Електронний ресурс].
  26. DISCO: Distributed Multi-domain SDN Controllers: Kevin Phemius, Mathieu Bouet and Jeremie Leguay [Електронний ресурс].
  27. FlowVisor: A Network Virtualization Layer: Deutsche Telekom Inc. R&D Lab, y Stanford University, _ Nicira Networks October 14, 2009 /Rob Sherwood, Glen Gibb, Kok-Kiong Yap [Електронний ресурс].
  28. Analysis of CAPEX and OPEX Benefits of Wireless Access Virtualization: the 4th Workshop on E2Nets, Budapest, Hungary, June 9, 2013 / M.M.Rahman, Charles Despins. [Електронний ресурс].
  29. Е.В. Дуравкин, Е.Б. Ткачева, Иссам Саад АРХИТЕКТУРА SDN. АНАЛИЗ ОСНОВНЫХ ПРОБЛЕМ НА ПУТИ РАЗВИТИЯ // Системи обробки інформації, 2015, випуск 3 (128).