With the increasing demand for data processing and storage in modern computing environments, innovations in data center networking have become essential. The shift toward cloud-based solutions and emerging paradigms like big data and IoT necessitate a re-evaluation of existing network architectures. Among various designs, the spine-leaf topology has gained popularity due to its scalability, low latency, and management simplicity. In a spine-leaf architecture, the backbone consists of central spine switches connected to leaf switches, enabling high-bandwidth, low-latency communication between servers. This structure ensures multiple paths between devices, helping to balance loads and reduce congestion. However, spine-leaf networks face challenges, including traffic congestion, resource underutilization, and fault tolerance. Our proposed solution, Dynamic Spine Scaling (DSS), leverages Software-Defined Networking (SDN) to dynamically adjust the number of active spine switches based on real-time traffic demands. DSS improves network adaptability, optimizing performance during peak loads and conserving energy during off-peak hours. To enhance accessibility for non-expert readers, we provide detailed SDN configuration parameters and setup information to facilitate reproducibility. Our study demonstrates significant performance improvements with DSS: latency reductions of up to 25% and energy savings of approximately 20% during low-demand periods. This work evaluates DSS within the spine-leaf architecture, assessing its impact on throughput, latency, fault tolerance, and energy efficiency.