Analysis and Comparison of Traditional Network and Software Defined Networking (SDN)

This thesis analyzes the architectural differences and performance characteristics of Traditional Networking (TN) and Software-Defined Networking (SDN), with a focus on latency (RTT), jitter, recovery time, and bandwidth enforcement capability. TN relies on distributed control and device-local configuration, while SDN separates the control and data planes to enable centralized, programmable management. To evaluate both paradigms under comparable conditions, TN was implemented in GNS3 and SDN was emulated in Mininet using a Ryu controller. Three representative topologies—star, linear, and tree—were deployed for testing. Performance measurements were obtained using ping for RTT, iperf3 for throughput and UDP jitter, and scripted link failures to quantify recovery time. A key objective of the study was to examine deterministic bandwidth control. In the SDN environment, priority-based policies allowed fixed capacity assignment per host and per traffic group, and measured throughputs closely matched the configured limits. By contrast, the TN configuration—lacking end-to-end QoS enforcement—could not guarantee a bounded throughput ceiling and behaved opportunistically under load. The results show that SDN consistently provides lower RTT, reduced jitter, and significantly faster recovery following failures due to centralized controller decision-making. Additionally, SDN enables fine-grained and enforceable bandwidth allocation, which was not achievable in the tested TN setup. Although the experiments were limited to simulated environments and a single controller, the findings demonstrate that SDN offers superior programmability, operational flexibility, and network control suited to modern deployment requirements.