Distributed Web Server’s Data Performance Processing with Application of Spatial Econometrics Models

Predicting the download time of Web distributed resources is an important but challenging problem in the worldwide public Internet. In our recent work we focus on spatial-based methods. We propose to use spatial econometrics methods to predict Web server’s performance. Three spatial regression models have been studied: Classical Regression Model (CRM), Spatial Lag Model (SLM) and Spatial Error Model (SEM). We used the real-life dataset obtained in active experiments performed by our Virtual Multiagent Internet Measurement System (VMWING), which monitored web transactions issued by VMWING’s agent located in Wroclaw, Poland and targeting Web servers in Europe. We also compared studied econometrics methods with geostatistical methods which were analyzed in our previous papers.