Model transformation chains and model management for end-to-end performance decision support
GTTSE'09 Proceedings of the 3rd international summer school conference on Generative and transformational techniques in software engineering III
Hi-index | 0.00 |
This paper presents a method for the complementation of a discrete event simulation with an analytical performance prediction approach to improve performance related decision support for business processes. This is enabled by the transformation of non-hierarchical tool independent performance models into hierarchical models of the analytical approach FMC-QE (Fundamental Modeling Concepts for Quantitative Evaluation), through a series of graph transformations. The method employs the so called Model-Driven Performance Engineering (MDPE) architecture. This architecture permits generation of input for different kinds of performance analysis tools, such as tools for discrete event simulations, via ``pushing-a-button''. While FMC-QE delivers a time efficient analytical performance evaluation mechanism with approximations for non-product form problems, discrete event simulations support models with arbitrary complex control flow structures at the cost of longer execution times. Therefore, we propose to employ the analytical approach for a sensitivity analysis, combined with simulations for detailed what-if questions and process optimizations. This approach is evaluated based on an industrial case study.