Implementation of Load Balancing Using the PCC (Per Connection Classifier) Method on Computer Networks to Improve Responsiveness

Abstract

Dependence on a single ISP can lead to the risk of internet disconnection, necessitating a solution to enhance network performance and responsiveness. This research explores the implementation of load balancing using the Per Connection Classifier (PCC) method to improve the responsiveness of computer networks utilizing three ISPs (Internet Service Providers). The test results indicate that the PCC method across the three ISPs can increase throughput by 9.32 Mbps, reduce delay by 43.54 ms, and decrease jitter by 74.17 ms while packet loss remains stable at 0.0%. Additionally, the recursive gateway failover technique in PCC load balancing demonstrates a significant increase in throughput up to 83.3 Mbps, a reduction in delay by 0.27 ms, and a reduction in jitter by 1.15 ms. Packet loss remains stable at 0.0% under varying ISP conditions. Moreover, a comparison between the PCC and PBR methods with the least connection algorithm shows that both methods effectively enhance network responsiveness, with the PCC method exhibiting superior performance in throughput.