Real-time UML-based performance engineering to aid manager's decisions in multi-project planning

We investigate the usage of software performance engineering to aid the project manager in making sound, reliable predictions in software development, and in optimizing the utilization of resources (typically the people). We assimilate the project teams to the processing elements of a performance model, and their activities to the tasks to be accomplished within established time intervals. The proposed methodology uses as the manager's interface a subset of Real-Time UML, the recently adopted OMG standard specialized profile for addressing schedulability, performance and timeliness. To illustrate the approach, we model the case of a manager that must decide a release date for a product undergoing the testing phase. We show how, by means of basic performance analyses, different workflow assumptions can be explored and their consequent outcomes automatically derived: by looking at the analysis results, the manager can thus take an informed, more responsible decision.