PERFORMANCE ANALYSIS OF LOAD BALANCING USING PER CONNECTION CLASSIFIER METHOD AND FUZZY METHOD ON NETWORKS WITH MULTI INTERNET CONNECTIONS

• Main Author: Risnaldy, Praja
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/86856
• Institution: Institut Teknologi Bandung

The internet is increasingly needed in various sectors. This is proven by the increasing number of users every year. However, this is not supported by network quality so that what happens in various sectors is that the Internet Service Provider (ISP) is down and the internet connection is slow. The process of repairing internet services that are down can take time so that it can cause losses for internet users. This can be overcome by using multiple internet connections to meet large internet needs by implementing load balancing. The implementation of load balancing aims to keep users connected when the connection is down and to improve slow internet connections. The implementation of load balancing is carried out using various methods. One method is Per Connection Classifier (PCC). This method uses hashing algorithm that works by ensuring that requests from the same user are always directed to the same gateway. This algorithm can cause the load on one of the connection paths to be higher. Therefore, a proposed load balancing method is needed to be compared with the PCC method, namely the Fuzzy method by implementing the fuzzy logic algorithm. The Fuzzy method works by calculating the amount of data to be sent and then selecting which gateway the data will be passed through optimally. Both load balancing methods are compared to determine the level of load distribution to each connection path, ISP 1 and ISP 2 by calculating the parameter values of throughput and delay. The results of the study show that the Fuzzy method load balancing is more even in distributing load compared to the PCC method as seen from the difference in the values of each parameter in the two ISPs, where the Fuzzy method has smaller throughput value difference difference between ISP 1 and ISP of 85.17% and smaller delay value difference between ISP 1 and ISP 2 of 71.43%.