Any model-based evaluation of the dependability of a system requires validation and verification to justify that its results are meaningful. Modern modeling frameworks like Mobius enable us to create and evaluate stochastic discrete event dynamic system models of great complexity. We believe that there is never enough support for an analyst in ensuring that the dynamic behavior of an executable model is consistent with the conceptual model that an analyst has in mind. In this talk, we describe ways to access information contained in simulation traces. We discuss a variety of features implemented in Traviando, a trace analyzer and visualizer. Traviando generates a report with observed model characteristics as a set of web pages. Furthermore, it provides a graphical user interface to support an interactive search for locations of interest in a lengthy simulation trace and a visualization of dynamic behavior with message sequence charts.
Peter Kemper is an Associate Professor in the Department of Computer Science at the College of William and Mary (having previously been at TU Dortmund and TU Dresden, Germany). His research interests include modeling techniques and tools for performance, performability, and dependability analysis of systems. He contributed to analysis techniques for the numerical analysis of Markov chains, model checking stochastic models, and techniques for simulation optimization.