Comparison Analysis of Load Balance Performance Per Connection Classifier (Pcc) And Equal Cost Multi-Path (Ecmp) Networks for Multiple Path Networks

Abstract

One technique to improve connection stability in a network is to use more than one Internet Service Provider (ISP). To make this possible, a load balancer is needed. Examples of load balancers are Per Connection Classifier (PCC) and Equal Cost Multi-Path (ECMP). However, both load balancers have drawbacks, where if there is an overload in one ISP line resulting in an imbalance in bandwidth usage, the two load balancers divert all traffic to another ISP. This study analyzes the performance of several load balancing techniques when applied to Mikrotik to maximize traffic flow with dual lanes. The experiment was carried out by applying the PCC, ECMP load balancer on the internet network. The independent parameters observed were throughput, jitter, and delay. The Fuzzy method will choose the ISP that will forward the data with the optimum bandwidth, because this method takes into account the amount of data that will be sent to the gateway through the ISP. The results of the measurement of Quality Of Service (QoS) network performance parameters such as throughput, jitter and delay with PCC and ECMP methods between ISP1 and ISP2 are throughput 22113.58 Kbps, Jitter 63.79 ms, delay 0.000578 ms. While in Utami's research (2017) with the PCC method, measuring Throughput is 2.9 Kbps, Delay is 135.25 ms and Jitter is 65.25 ms.