Discussing the Availability Quality Attribute in Systems-of-Systems Architectures based on a Simulation Experiment

Systems-of-Systems (SoS) is a set of independent software-intensive systems intended to support critical domains, such as emergency and crisis response systems and health. As such, SoS should exhibit robust behaviors and guarantee that the entire system is continuously available, not failing and leaving its users helpless. However, the SoS inherent dynamic architecture can potentially affect the availability of the functionalities being offered by the SoS. Hence, it is prominently important to anticipate, at design-time, the degree of availability an SoS can offer and provide mechanisms to reinforce the availability provided by it. Usually, software architects use static models to predict quality attributes in software architectures. However, these models are not well-succeeded to deal with the dynamic nature of the SoS architectures. The main contribution of this paper is thus presenting emerging results on the establishment of a simulation-based approach for empirically predicting the availability of SoS software architectures at design time. A Flood Monitoring SoS was used as the experimental platform. Preliminary results indicate that, specifically for the context of our study, the nature of availability in SoS architectures was not only influenced by the availability of the SoS constituents, but also by the SoS topology.