Integrating Performance Analysis in Model Driven Software Product Line Engineering

A Software Product Line (SPL) is a set of similar software systems that share a common set of features satisfying a particular domain, and are built from a shared set of software assets using a common means of production. This research proposes to integrate performance analysis in the early phases of the model-driven development process of SPL. We start by adding generic performance annotations to the UML model representing the set of core reusable SPL assets using the MARTE Profile adopted by OMG. A model transformation realized in Atlas Transformation Language (ATL), derives the UML model of a specific product with concrete performance annotations from the SPL model, which is further transformed into a performance model by using a previously developed transformation called PUMA.;Performance is a run-time property of the deployed system and depends on other factors that are external to the design model, characterizing the underlying platforms and run-time environment. Performance completions provide a means to extend the modeling constructs of a system by including the influence of these factors. The variability space of the performance completions is covered and represented through Performance Completion-feature model (PC-feature model).;Dealing manually with a large number of performance parameters annotating a UML+MARTE product model is an error-prone process. A model-driven user-friendly technique is proposed to automatically collect all generic performance parameters that need binding from the generated product model and present them to developers in a spreadsheet format, together with context and guiding information where each PC-feature is mapped to certain MARTE annotations corresponding to UML model elements in the product model.;The automatic derivation of a specific product model based on a given feature configuration is enabled through the mapping between features from the feature model and their realizations in the design model. An efficient mapping technique is proposed that aims to minimize the amount of explicit feature annotations in the UML model of SPL. Implicit feature mapping is inferred during product derivation from the relationships between annotated and non-annotated model elements as defined in the UML metamodel. The mapping technique is used to derive automatically a given product model.