Peningkatan Efisiensi Penggunaan Software Defined Network Menggunakan Metode Least Load Path

Load imbalance is a major problem in Software Defined Network (SDN) architectures, occurring when traffic distribution is uneven so that some links experience congestion while other paths remain underutilized. This study aims to implement the Least Load Path (LLP) method to improve the utilization efficiency of SDN and to compare it with the Shortest Path (SP) and Simple Switch (SS) methods. The study was conducted using the Mininet emulator with a multipath topology of seven OpenFlow switches, the Ryu Controller, and NetworkX. The LLP method monitors OpenFlow port statistics every five seconds, computes the load of each link using the formula Load (bps) = (Δbyte × 8) / time interval as edge weights, and applies Dijkstra's algorithm to find the path with the lowest total weight. Testing was carried out under three load scenarios, measuring packet loss, Round Trip Time (RTT), throughput, and path selection. The results show that LLP maintained 0% packet loss across all scenarios, preserved an average RTT of 27.24 ms under high load (95.84% better than SP which degraded to 655.64 ms), and achieved a throughput of 72.30 Mbps under medium load (36.16% higher than SP and 43.7% higher than SS). LLP was also the only method capable of dynamic path selection. This study confirms that LLP is an effective and adaptive load balancing mechanism for SDN environments.