Performance Evaluation of Distributed Web Server Architectures under E-Commerce Workloads

One of the central and key components to make the ECommerce a success is high performance and highly reliable web server architectures. Extensive research has been sparked aiming at improving web server performance. Fundamental to the goal of improving web server performance is a solid understanding of behavior and performance of web servers. However, very little research is found on evaluating web server performance based on realistic workload representing E-Commerce applications which are usually exemplified by a large amount of CGI (Common Gateway Interface), ASP (Active Server Page), or Servlet (Java Server-side interface) calls. This paper presents a performance study under the workload with a mixture of static web page requests, CGI requests, Servlet requests, and database queries. System throughputs and user response times are measured for five different server architectures consisting of PCs that run both a web server program and a database. We observed that performance behaviors of the web server architectures considered under this mixed workload are quite different from that under static page workload and sometimes counter-intuitive. Our performance results suggest that there is a large room for potential performance improvement for web servers.