Distribution of control using active storage

Abstract

Real-time control systems are often distributed, with complex data-sharing requirements. A method is proposed for the distribution of functions in such a control system using a distributed real-time database as storage, extended with an active mechanism based on event–condition–action rules that react to changes in the system. This database is fully replicated and nodes communicate using only the database distribution facilities. Critical temporal constraints at individual nodes may be met at the expense of introducing temporary inconsistencies in the database. This paper investigates the effects of using the active mechanism and the replicated storage as a distribution mechanism for certain classes of real-time systems. Particular attention is paid to the requirement of applications to tolerate temporary inconsistencies, assuming that eventual consistency is ensured.