Customisable model transformations based on non-functional requirements

Abstract

The Model-to-Model (M2M) transformation stage in a Model-Driven Engineering (MDE) tool chain is used to bridge the large semantic gap between problem-domain abstractions and software artefacts. This transformation is typically specified in a closed, monolithic way. Since no two systems have identical requirements, some work has been done to create explicit support for customisation of the transformation process based on functional requirements. The same is not true for non-functional requirements. The contribution of this paper is a customizable M2M transformation process that takes as an input the users priorities, in terms of non-functional properties, to select between design trade-offs in the transformation. These tradeoffs are expressed as contrasting architectural styles, which describe common patterns of interactions and constraints with well understood non-functional implications. The target of the M2M transformation is an Architecture Description Language (ADL) that can be used to express styles. Model checking tools can be used to ensure that styles are adhered to during the entire MDE process. This early work focuses on describing contrasting architectural styles that address distribution issues, such as fault tolerance.