Abstract: Many software-intensive systems today can be characterized as systems of systems (SoS) comprising complex, interrelated, and heterogeneous systems. The behavior of SoS often only emerges at runtime due to complex interactions between the involved systems and their environment. It is thus necessary to determine unexpected behavior by monitoring SoS at runtime, i.e., collecting and analyzing events and data at different layers and levels of granularity. Existing monitoring approaches are often limited to individual systems, particular architectural styles, or technologies. In this paper we thus derive challenges for monitoring SoS based on an industrial case. We present a flexible framework adaptable to different system architectures and technologies. We discuss its capabilities for instrumenting systems, collecting and persisting events and data, checking constraints on events and data, and visualizing the systems’ behavior to users. We demonstrate the framework’s flexibility by tailoring and applying it to an industrial SoS and assessing its performance and scalability. Our results show that the framework is flexible and scalable for monitoring an industrial SoS with realistic event loads.
Authors: Michael Vierhauser, Rick Rabiser, Paul Grünbacher, Klaus Seyerlehner, Stefan Wallner, Helmut Zeisel