Integrating temporal, real-time, an active databases

Abstract

To meet the needs of many real-world control applications, concepts from Temporal, Real-Time, and Active Databases must be integrated: Since the system's data is supposed to reflect the environment being controlled, they must be updated frequently to maintain temporal validity; Many activities, including those that perform the updates, work under time constraints; The occurrence of events, for example, emergency events, trigger actions. In these systems, meeting timeliness, predictability, and QoS guarantee requirements — through appropriate resource and overload management — become very important. So, algorithms and protocols for concurrency control, recovery, and scheduling are needed. These algorithms must exploit semantics of the data and the transactions to be responsive and efficient. Whereas time cognizant scheduling, concurrency control and conflict resolution have been studied in the literature, recovery issues have not. We have developed strategies for data placement at the appropriate level of memory hierarchy, for avoiding undoing/redoing by exploiting data/transaction characteristics, and for placing logs at the appropriate level in the memory hierarchy. Another issue that we have studied deals with the assignment of priority to transactions in active real-time database systems. We are also studying concurrency control for temporal and multi-media data. We have built RADEx, a simulation environment to evaluate our solutions.