Promising methods to improve the performance of the software-defined networking controller
MetadataShow full item record
With the continuous development of computer networking, a flexible network architecture will likely have to adapt to the recent growth of Internet-based systems, such as Cloud-based systems, the Internet of Things (IoT), and the Fifth-Generation (5G) cellular networks. Software- Defined Networking (SDN) is a new paradigm that simplifies the organization of data communications, facilitates the evolution of computer networking, and paves the way to absorb the potential requirements of future network changes. SDN aims to decouple the control function from the end network devices (i.e., routers) and provide an external centralized entity for all the network's control activities. However, currently the SDN's control capabilities are limited to the performance of a single controller. Our research aims to provide potential solutions addressing SDN's control limitations. To improve network performance, we introduce several models to address the interaction between the SDN controller and the network switches. The core contribution of this research includes the introduction of assistant switches, edge controllers, and self-routing traffic flows. The goal of the introduced models is to alleviate the controller's burden and improve its processing efficiency. To demonstrate the effectiveness of the proposed techniques, we implement a simulated SDN-based network using Mininet and the NOX network software platform with one controller connected to 10-32 switches that carry and route traffic flows among multiple end-hosts. The results of the simulation studies show a significant improvement in performance, including more than a 30% decrease in the amount of data transmitted by the SDN controller, up to 45% decrease in bandwidth usage, and up to 29% decrease in controller response time. The proposed methods can be extended to explore and develop more methods of collaboration to address some of the major issues associated with the centralized controlling architecture.
Thesis (Ph.D.)-- Wichita State University, College of Engineering, Dept. of Electrical Engineering & Computer Science