KLAPER: an intermediate language for model-driven predictive analysis of performance and reliability

Automatic prediction tools play a key role in enabling the use of non-functional analysis results in the selection and assembly of components for component-based systems, without requiring extensive knowledge of analysis methodologies to the application designer. To achieve this goal, a key idea is to define a model transformation that takes as input some “design-oriented” model of the component assembly and produces as a result an “analysis-oriented” model that lends itself to the application of some analysis methodology. For this purpose, we define a model-driven transformation framework, centered around a kernel language whose aim is to capture the relevant information for the analysis of non-functional attributes of component-based systems, with a focus on performance and reliability. Using this kernel language as a bridge between design-oriented and analysis-oriented notations we reduce the burden of defining a variety of direct transformations from the former to the latter to the less complex problem of defining transformations to/from the kernel language. The proposed kernel language is defined within the MOF (Meta-Object Facility) framework, to allow the exploitation of existing model transformation facilities. In this chapter, we present the key concepts of our methodology and we show its application to the CoCoME case study.