Using ATL for transformations in software performance engineering: a step ahead of java-based transformations?

Abstract

Transformations of software models (such as UML diagrams) into non-functional models (such as Queueing Networks) have brought a real breakthrough to the entire field of non-functional software validation, because they allow to introduce automatism in the generation of a non-functional model from software artifacts. However, up today almost all the existing approaches are based on general purpose programming languages, such as Java. With the rapid evolution of model transformation languages, it is interesting to study how transformations in the software performance engineering domain may benefit from using constructs and tools of these languages. In this paper we present the results of our implementation, in ATLAS Transformation Language (ATL), of a transformation approach from UML models to Queueing Network models and, laying on a previous implementation of the same transformation in Java, we discuss the differences between these two approaches.