Load Balancing Analysis Using Round-Robin and Least-Connection Algorithms for Server Service Response Time

Abstract

Today's internet network is expanding quickly, which encourages consumers to connect to more server services. A reliable server system is required to handle these circumstances. One approach that can be taken is to implement multiple servers. However, using many servers affects the response time of the server to be able to serve requests. Improving server system services through the load balancing method needs to be balanced with analysis, testing and evaluation of the network architecture and algorithms used in order to produce an optimal server system. In order to determine the algorithm to be tested optimal response time value, this study compares and analyses the load balancing approach utilizing the least-connection algorithm and the round-robin algorithm. Based on the test results using a request value of 500 connections/second for 1000 requests and 600 connections/second for 1200 requests, the round-robin algorithm looks slightly better than the least-connection algorithm. However, in the test scenario for the value of 700 connections/second for 1400 requests, 800 connections/second for 1600 requests, and 900 connections/second for 1800 requests, the least-connection algorithm looks superior. So that the average response time of the least-connection algorithm is smaller when there is an increase in connections. This condition has an impact on the faster the server responds to requests from users so that server performance can be continuously improved.