Use case-based evaluation of software defined networking controllers

Thumbnail Image
Journal Title
Journal ISSN
Volume Title
University of Zululand
The paradigm of Software Defined Networking (SDN) model has arisen as a viable candidate to tackle traffic management and costs based problems in developing markets. At the root of SDN is the control system that responsible for overseeing resource orchestration premised around its global understanding of the existing resource utilization of the network. A significant problem that needs to be resolved in the early implementation of an SDN-based network configuration is where each controller and switches are located to ensure the optimal operational performance. It is critical to note that the placement and size of the data plane devices have a direct impact on the overall performance of the SDN controller. This problem represents conflicting priorities, such as delay, dynamic routing and efficiency. In order to address the controller performance problem and formulate a strategy for the deployment of SDN in emerging markets, the study proposes solutions that provide flexibility, ease of management, traffic engineering and network distribution with the aim of reducing delay and of enabling efficient network utilization. There are important questions that should be answered during transitioning to an SDN solution, and those questions are: Given a legacy WAN, how many SDN controllers and switches are required? How should they be placed? Where are the best locations to place them? And how efficient are they in terms of traffic engineering so that the overall network performance is optimized? Henceforth, we will refer to these questions as the switch-tocontroller performance problem. The average latency was reduced to 15% compared to that of the Optimized K-means. The proposed method was also successful in the throughput (with an improvement of 13%). The studied metrics on RTT and throughput showed improvement in the reliability test equal to averages of 1.182 and 2.166 ms with two and three-controller deployment respectively and a throughput average of 20 GB/s. The results have displayed a positive impact in terms of improved controller performance on the selected use-cases. The tree topology yielded an improved average RTT of 0.4 ms and 0.2 to 0.3. GB/s in transfer speed and 0.2 Gbps in bandwidth utilization in Section 6.2.1. The ring topology resulted in an average of 0.6 ms RTT, the transfer speed of 1.2 GB/s and bandwidth of 0.2 Gbps in Section 6.2.2. The mesh topology resulted in an average RTT test of 0.5 ms, with a transfer speed of 0.9 GB/s and a 0.1 Gbps of bandwidth utilization in Section 6.2.3. The Torus topology in RTT yielded an average of 0.6 ms in Section 6.2.4. The transfer speed improved by 0.1 GB/s and bandwidth utilization was not very efficient at 0.2 GB/s. This study can be used by network engineers as a foundation to start integrating SDN or plan a new SDN deployment, by assisting engineers to make quick decisions about optimal switch placement and SDN deployment.
• A dissertation submitted in fulfillment of the academic requirements for the degree of Master of Science in the Department of Computer Science in the Faculty of Science, Agriculture and Engineering, University of Zululand, 2020.
Case-Based Evaluation, Networking Controllers