ABSTRACT
The increasing number of real-time applications, and their intrinsic need to manage larger amounts of data, present the system designer with software architectural challenges at the time of development and through the system life cycle. Real-time database research has successfully shown the merits of integrating database support for dealing with data as a resource, because a data-centric view enables simpler capturing of real-world data requirements; e.g., temporal validity, consistency, and quality, without losing semantics under a strict task-centric approach. Each application imposes requirements on the architecture with respect to the combination of memory consumption, processing capacity, hardware and operating system platform, data quality requirements, workload arrival patterns, stringency and urgency of timeliness, etc. These requirements imply that common support for dealing with real-time data efficiently needs to adhere to a unique combination of requirements as it must be tailored for the application. The incentives for developing a real-time database platform that can be specifically tailored to a range of applications are (i) economic as software is reused and development time decreases, (ii) simplified maintainability as a common database platform can be used across a family of real-time systems, and (iii) easier modeling of data requirements.
