Towards performance-oriented perfective evolution of BPMN models

Simulation techniques are successfully applied to analyze and validate the performance of a business process (BP) since the early phases of its lifecycle, when the BP representation is commonly specified in BPMN. The BP simulation model is first to be built from the BPMN model, then implemented and finally executed to yield the performance indices of interest. The model building activity includes a parameterization task that is carried out either by use of tool-specific facilities or through standardized languages for specifying simulation model parameters. In both cases, the parameters have to be externally linked to the BPMN model. A similar reasoning applies to the simulation results. This paper introduces a BPMN extension to annotate BPMN models with both the input parameters and the results provided by the BP analysis, so as to include in a single BPMN model all the data associated to the simulation-based analysis of a given BP. The paper also outlines an architecture that exploits the proposed extension to realize the performance-oriented perfective evolution of BPMN models, or the ability to introduce a feedback chain that makes use of simulation results to automatically refactor a BP model in order to better meet the BP performance objectives.