As datacenters evolve to prioritize dynamic and high-through put workloads, traditional network architectures face challenges in scalability, flexibility, and automation. This thesis compares traditional networking with Software-Defined Networking(SDN) using spine-leaf topologies to evaluate performance and efficiency in a datacenter context. A physical topology in a lab was used to compare a traditional network with a Mininet-emulated SDN topology using Ryu as the controller. Performance was evaluated through idle packet count, convergence time, and packet retransmission during link failures. Results showed that SDN generated significantly less control traffic and converged faster than traditional setups; the traditional network showed no packet loss under link failure, while the SDN did. Operational tasks such as monitoring and scalability were also compared. The comparison suggests that SDN offers more scalable and centralized management capabilities, streamlining configuration and monitoring processes. The SDN demonstrates clear advantages in programmability leading to better scalability and monitoring ,indicating its growing suitability for modern data centers.