REMINDS: Requirements Monitoring for Systems of Systems -- Princeton
Sprache des Vortragstitels:
Englisch
Original Kurzfassung:
Many software-intensive systems today are very-large-scale software systems with systems of systems (SoS) architectures comprising interrelated and heterogeneous systems developed by diverse teams over many years. Due to their scale, complexity, heterogeneity, and variability engineers face significant challenges when determining the compliance of SoS with their requirements. In particular, certain behavior only emerges at runtime due to complex interactions between the involved systems and their environment. Monitoring the behavior of SoS at runtime is thus essential. However, existing requirements monitoring approaches are often limited to particular architectural styles or technologies and are thus hard to apply in SoS architectures. They do not adequately consider the characteristics of SoS: requirements exist at different levels, across different systems, and are owned by diverse stakeholders. This talk provides an overview of the research on requirements monitoring for very-large-scale software systems conducted at the Christian Doppler Laboratory MEVSS at the Johannes Kepler University Linz, Austria. More specifically, in cooperation with Primetals Technologies, we have been developing REMINDS, a flexible framework for runtime monitoring of system-of-systems architectures, which is based on a requirements monitoring model defining the key elements to be monitored and their relations: requirements in SoS can have different scopes and have to be refined and formalized as constraints to allow checking them at runtime. The SoS is instrumented using probes, which provide events and event data at runtime to a unified event model. Constraints operate on this model and check events and attached data. The separation of concerns between the actual systems and a higher-level instance for constraint definition and evaluation allows the definition of cross-cutting and global constraints, which require data aggregated from various systems.