Spatio-temporal Web Performance Prediction: Turning Bands Method and Sequential Gaussian Simulation

Abstract

This paper presents a novel geostatistical approach in Web engineering that allows not only to evaluate but also to predict Web applications performance. The approach presented in this article can be used for two purposes: the first, to find (or forecast) which of the considered web servers are running web applications in the most efficient way and the second, to check (or forecast) if an evaluated web software meets given performance criteria. The first part of this paper briefly elucidates two geostatistical methods used in this research: the Turning Bands Method and Sequential Gaussian Simulation. The second part characterises the multiagent system MWING, a software solution that by conducting active measurements collects data necessary for analysing, evaluating and forecasting Web applications performance. The final part presents a case study of web performance prediction approaches proposed by the authors.