Web Delay Analysis and Reduction by Use of Load Balancing of a Dispatcher-based Web Server Cluster

The loading of web servers can be very heavy and a delay in accessing certain Web sites can be substantial. Hence, the reduction of web delay is one of the important topics of research for improving the quality of service of Web pages. Thus, in this paper, we propose the web delay analysis and the web delay reduction by using load balancing coupled with the dispatcher-based approach. First, we propose a queueing model to represent the web delay. Second, we propose a mathematical model to obtain the least possible web delay by the use of load balancing method. By using the first and second portions as simulated by QNAT, we find the cluster of web servers may not have incorporated any load balancing for all situations. Thus, we propose two strategies to justify the existence of load balancing. The first one is a simulation module that we design by using Virtual C++. The second is a Hessian matrix improve the computation efficiency of web delay. In addition, to simplify our analysis, we produce a pre-compute table in order that people may judge the load balancing of web servers without using complicated computation. Overall, we learn that load balancing will reduce the web delay substantially. Moreover, by meeting the specific conditions of the load balancing scheme, we can have a potential to obtain a web delay reduction of 38.46%.