Performance analysis roundtrip: automatic generation of performance models and results feedback using cross-model trace links
Taghreed Altamimi, Mana Hassanzadeh Zargari, Dorina Corina Petriu · Computer Science and Software Engineering · 2016
This paper proposes an approach for performance analysis roundtrip in the context of model-driven engineering (MDE) of real-time distributed and embedded systems. The starting point is a UML software model with MARTE performance annotations, such as performance requirements and resource demands. The source software model is automatically transformed into a Layered Queueing Network (LQN) performance model. We developed the transformation with Epsilon, a family of languages for model-to-model transformation, model validation and model management. Using specialized languages helped us create a more compact transformation, easier to understand and maintain than transformations developed with general purpose languages, such as Java. Beside the performance model, the transformation also generates a traceability model containing trace links between mapped elements of the software and performance model. After solving the performance model with an existing solver, the performance results are fed back to the software model by following in reverse the cross-model trace links. The software developers can see the performance results as MARTE stereotype attributes, using a standard UML editor. The approach is illustrated by applying it to an e-commerce application.