Simulation-Based Performance Evaluation of Software Architectures described in UML

Quantitative analysis of software systems is being recognized as an important part of the software development process. In particular, performance analysis can be particularly helpful in identifying possible bottlenecks during the early stages of the software development life cycle. Early identification of problems is particularly desirable as the costs of changing the design are much higher the latter in the design cycle the change is done. In this paper we will focus on simulation-oriented performance models of software systems. We propose a methodology for translating a UML design of a software architecture into a simulation model, which can then be implemented in a discrete-event simulation program. The UML model of the system is annotated using the UML profile for Schedulability, Performance and Time. This allows quantitative informations to be easily added to the model in a standard and consistent way.