Design and implementation of SDN fault tolerant framework and security services on-demand

Abstract

Software-defined networking (SDN) revolutionizes network management by separating control and data planes, improving programmability, and automating operations. However, the centralized SDN control plane which houses the network intelligence presents challenges such as single points of failure with single controllers or added complexity with multi-controllers. Despite these challenges, SDN offers critical functionalities for security and on-demand services. This thesis addresses SDN's challenges by focusing on fault tolerance and security management. The study proposes a fault-tolerance model (FTM) to improve SDN reliability and applies it to security management. The FTM prioritizes fault monitoring, detection, and recovery as well as ensuring scalability and performance across networks of varying sizes: small to large-scale networks. By utilizing design science research methodology, this study designed an FTM which employs backup controllers and stationary agents for small networks, while introducing a novel controller placement technique for large networks. In addition, an on-demand security service framework that leverages SDN functionalities to improve network protection is designed. Performance evaluation in Mininet simulations based on OpenFlow and Ryu SDN controllers using different evaluation metrics demonstrates the effectiveness of the proposed FTM, with promising performance in certain scenarios. Particularly, the OpenFlow controllers exhibit lower latency rates after failure in small-scale networks, while Ryu controllers outperform in medium to large-scale networks in terms of the number of controllers deployed and network availability. These findings provide customized implementation strategies for different network sizes and recommend the adoption of the frameworks in real-world SDN environments to enhance network resiliency and dependability