Plenary Lecture

Modelling the Performance Behavior of a Web Server using Statistical Factor Analysis

Assistant Professor Livia Sangeorzan
Transilvania University of Brasov
Member of the Applied Computer Science Department,
Research Laboratory Coordinator
Romania
E-mail: livia.sangeorzan@gmail.com

Abstract: In studying the time in executing the programs on the Web server, on can made the analysis of variance technique in a more complex design, on can made a factorial design.
On can fixed the following tuple: (architecture, no. of processes, JVM state, application, heap size, no threads, execution time).
Analysis of variance does not ignore interaction. In more complex statistical designs estimations of variance due to interaction are made.
In the analysis of variance techniques all of the data are treated at once and a general null hypothesis of no difference among the means of the various groups is tested.
It is desirable to describe the joint relationship of a single Y variable to several X variables. If the Y variable is well described by the other variables we will want to know the existent of this dependence and the actual multiple regression equation. On can determinate some correlations between the fixed tuple.
Using the multiple regression equation determinate, on can modelling the performance of a Web server behavior.