Date of Award

2005

Publication Type

Master Thesis

Degree Name

M.Sc.

Department

Computer Science

Keywords

Computer Science.

Rights

info:eu-repo/semantics/openAccess

Abstract

By the time the architecture of a software system is decided in the life-cycle of software development, performance problems become very costly, if not impossible, to fix. It is, therefore, necessary to push performance analysis back to the architectural design stage as an effective means to improve the performance of software systems. This is especially true for web services-based systems, where system performance is of paramount importance. There are typically three steps in performance evaluation of software architectures. The first step is to transform the architecture of a software system in forms of annotated UML models into a performance model, such as the layered queueing network model (LQN). Experiments on the performance model are then conducted in the second step with a performance analysis tool, such as the LQN solver. Experiment results are finally fed back to architecture design in the last step for refinement of UML models according to the quantitative analysis of software performance. Nevertheless, accurate analysis results require performance analysis to take sensitivity analysis into consideration in between the second and third steps. Unfortunately, little research has been done in this regard. This thesis carries out a study in performance analysis with sensitivity analysis. It develops a new method that uses the design of experiments (DoE) techniques to quantitatively analyze the sensitivity of a system's performance output due to the effect of the system's input factors, and the interaction between those factors. The goal of this research is to provide more accurate feedback to software designers on the development of service-oriented software systems. Paper copy at Leddy Library: Theses & Major Papers - Basement, West Bldg. / Call Number: Thesis2004 .H8245. Source: Masters Abstracts International, Volume: 44-01, page: 0393. Thesis (M.Sc.)--University of Windsor (Canada), 2005.

Download

Share

COinS