Comparative Analysis of QoS of Load Balancing Network Using Equal Cost Multi Path And Per Connection Classifier Methods

Abstract

The rapid development of information and communication technology demands the availability of fast, stable, and reliable internet access. However, a common issue that arises is slow internet performance caused by high user traffic and suboptimal network management. This study aims to analyze and compare the Quality of Service (QoS) of load balancing using the Equal Cost Multi Path (ECMP) and Per Connection Classifier (PCC) methods, with a case study conducted at the Engineering Faculty Laboratory of Universitas Khairun. The research method applies the Network Development Life Cycle (NDLC), consisting of analysis, design, simulation, implementation, and testing stages. The devices used include a Mikrotik RB450GX4 router, two Internet Service Providers (Astinet and Indihome), and other supporting equipment. The implementation involves basic configuration, NAT, mangle, routing, and failover setup. QoS testing was carried out based on parameters such as throughput, delay, jitter, and traffic distribution stability between ISPs. The results show that both load balancing methods can improve connection stability and effectively distribute network traffic between ISPs. The ECMP method distributes data packets alternately across available paths, while the PCC method performs better in maintaining connection consistency based on address and port.From the testing results, the PCC method achieved a more balanced traffic distribution but may lead to overload if multiple connections target the same destination simultaneously. Meanwhile, the ECMP method is simpler and more effective in maintaining network traffic with good failover support. Overall, implementing load balancing using ECMP and PCC methods has proven to enhance the network service quality. This study is expected to serve as a reference for managing multi-ISP networks to support academic and operational activities.