The most critical requirements for the lifetime value of a system are its nonfunctionalrnrequirements such as reliability, security, maintainability, changeability, etc. Thesernare collectively known as the ‘ilities’ and are typically addressed in system design once thernfunctional architecture has been developed. In this paper we propose the use of architecturecentricrndesign that modifies this standard workflow so that those non-functionalrnrequirements, which actually reflect the true business needs, are addressed first. This ensuresrnthat the final system better reflects and embodies those architecturally-significantrnrequirements rather than having them addressed secondarily. This is an important changernsince the ‘ilities’ are systemic properties (properties of the system as a whole) rather thanrnsystematic properties (properties of individual components or sub-systems) and are thereforerndifficult to address once the functional architecture has been determined and the separation ofrnconcerns is already somewhat completed. We provide an example of the approach basedrnaround a simplified case study of an online, on-board health monitoring system for shipboardrngas turbine electricity generators that collects, filters, analyzes, transmits, and mines sensorrndata from the generators subsystems.
展开▼