Content-Based Load Balancing Using Web Clusters and Service Rankings

Abstract

Web servers have gained immense popularity currently due to its nature to cater to a huge number of user requests, where any number of users can get service from Web-based applications. As it comes with global accessibility, servers hosting popular websites tend to get massive load that deteriorates in efficiency to provide quality service. Currently, most servers use load balancing techniques that distribute load among multiple virtual servers hosting a website. Generally, these load balancing techniques concentrate on balancing user request load on server and ignore the type of content requested. The reason behind it is overhead on dispatcher for analyzing content request and assigning appropriate rank. This process consumes additional time for dispatcher to allocate server and users face little delay in viewing their site; however it is helpful in the long run as heavy bandwidth consuming services such as multimedia requests can be catered efficiently by high priority servers only. The priority of servers can be defined as per their configuration and capabilities. Through experimental results on J Meter software it is proved that this concept can be helpful in providing better Quality of Service (QOS) to the website users.