Configurable Software Performance Completions through Higher-Order Model Transformations

Abstract

Chillies is a novel approach for variable model transformations closing the gap between abstract architecture models, used for performance prediction, and required low-level details. We enable variability of transformations using chain of generators based on the Higher-Order Transformation (HOT). HOTs target different goals, such as template instantiation or transformation composition. In addition, we discuss state-dependent behavior in prediction models and quality of model transformations.