Empirical Evaluation of Model-Based Performance Prediction Methods in Software Development

Predicting the performance of software architectures during early design stages is an active field of research in software engineering. It is expected that accurate predictions minimize the risk of performance problems in software systems by a great extent. This would improve quality and save development time and costs of subsequent code fixings. Although a lot of different methods have been proposed, none of them have gained widespread application in practice. In this paper we describe the evaluation and comparison of three approaches for early performance predictions (Software Performance Engineering (SPE), Capacity Planning (CP) and umlPSI). We conducted an experiment with 31 computer science students. Our results show that SPE and CP are suited for supporting performance design decisions in our scenario. CP is also able to validate performance goals as stated in requirement documents under certain conditions. We found that SPE and CP are matured, yet lack the proper tool support that would ease their application in practice.