A feasible approach to predict the response time for 3-tier web application with service demand law

Abstract

The web applications have grown rapidly and become ubiquitous. This paper proposes a simple and feasible approach to predict one Quality of Service (QoS) metric of typical 3-tier web application, the average response time, which plays a crucial role in maintaining customers' loyalty. Our approach is sufficiently general to be used in various web applications, we use Service Demand Law to estimate service demand of each resource and use Queuing Network theory to model the application as a network of resources queues. We validate our model in a 3-tier enterprise e-commerce application test bed by comparing the average response times that predicted by our model and that observed in the actual measurements. The validation results yield that our approach is effective for predicting the average response time for 3-tier web application. © Springer-Verlag Berlin Heidelberg 2012.