Resource Contention Analysis of Cloud-based System through fUML-driven Model Execution.
Martin Fleck, Luca Berardinelli, Philip Langer, Tanja Mayerhofer, Vittorio Cortellessa · 2013
Model-driven software engineering not only enables the efficient development of software but also facilitates the analysis of non-functional properties (NFPs). As UML, the most adopted modeling language for designing software, lacks in formal execution semantics, current approaches translate UML models into dedicated analysis models, before NFPs can be computed. However, such transformations introduce additional complexity for the users and developers of analysis tools. To avoid this additional complexity, we show how the analysis of certain NFPs can be performed solely based on UML models without translating them into, e.g., queuing networks. Therefore, we leverage the execution semantics of fUML, a recent OMG standard, to gain execution traces from UML models and, based on these traces, compute indices, such as response time, taking into account the contention of resources, as well as different resource management configurations, such as balancing and scaling strategies.