Failure localization in SDN networks

OData support
Supervisor:
Pasic Alija
Department of Telecommunications and Media Informatics

The development of data transport networks is a current an important research area. Within the field the Software Defined Networking (SDN) is a promising new technology. In the future, this technology could potentially solve numerous problems encountered in the current hardware based systems. Due to its design and operation, SDN eases the innovation, like the developement in failure localization methods.

There are algorithms available for single link failure, however real large scale networks often have double or multiple link failure. The purpose of the thesis is to design and implement an algorithm which makes it possible to monitor a dual link failure. I have tried to create an algorithm, which uses the opportunity given by SDN, and can provide solution for the failure localization of SDN networks.

The most novel improvement is a graph theory algorithm, the use of ear decomposition. The location of two, for larger networks even more link failures can be determined during the algorithm operation.

The principles of SDN and failure localization is introduced in the thesis. Two already available failure localization solutions are presented. One is the Link Layer Discovery Protocol, which is mainly used in the hardware-based networks, and the other failure localization algorithm is the SDN specific solution, which is based on the monitoring route. In addition to presenting the theoretical principles I also demonstrate the operation of the algorithm I have created. The technologies used to implement the algorithm are detailed, and the operation of the executing software is presented. I process the test operation results, and highlight the possible improvement of the algorithm.

The algorithm, I have created and its implementation meets the requirements, it can be a good solution for failure localization in SDN based networks.

Downloads

Please sign in to download the files of this thesis.