Pengujian Distribusi Beban Kerja Web Secara Statis pada Sistem Server Web Berbasis Cluster dengan Algoritma Never Queue

Abstract

The increase in web traffic and the development of network bandwidth that is relatively faster than the development of microprocessor technology today causes the one point server platform to be no longer sufficient to meet the scalability requirements of web server systems. Multiple server platforms are the answer. One known solution is cluster-based web server systems. In this study, a cluster-based web server system would be designed with the Never Queue algorithm and continued with testing the distribution of web workload on this system. The tests were carried out by generating HTTP workloads statically (with fast HTTP requests per fixed second) and dynamically (rapid HTTP requests per second that change or rise regularly) from the client to the web server system pool. Followed by analyzing data package traffic. In this study, the results of static testing with rapid HTTP requests per second which still showed that the Never Queue algorithm distributed HTTP requests to the web server system pool properly and got HTTP replies that tend to be stable at the HTTP average of 1031.8 replies/s. As for the rapid parameters of TCP connections, response times and errors increased with the rapid increasing HTTP requests generated. The average output was at 2,983 Mbps.