The need for stable and efficient internet in the small business environment and public services demands optimal traffic distribution. Conventional load-balancing methods such as PCC and ECMP are often unable to allocate bandwidth based on service type or application priority, leading to congestion and decreased performance during peak hours. This study developed a load-balancing system based on Policy-Based Routing (PBR) to optimize traffic on multi-ISP networks using MikroTik RouterOS. The system is configured by classifying traffic based on the source IP, protocol type, and service needs, then routed through two ISPs using separate routing tables. The test results showed that PBR significantly improved network performance, with a 20–35% increase in throughput, a 60% decrease in latency, and a more stable QoS in streaming, VoIP, and browsing services. This study proves that PBR provides greater flexibility and efficiency than conventional load-balancing methods, making it suitable for applications in cafes, MSMEs, and agencies that require stable multi-link connectivity.
Copyrights © 2025