Commit processing in distributed real-time database systems

Incorporating distributed data into the real-time framework incurs the well-known additional complexities that are associated with transaction concurrency control and database recovery in distributed database systems. We investigate here the performance implications of supporting transaction atomicity in a distributed real-time database system. Using a detailed simulation model of a rm-deadline distributed real-time database system, we pro le the real-time performance of a representative set of commit protocols. A new commit protocol that is designed for the real-time domain and allows transactions to \optimistically" read uncommitted data is also proposed and evaluated. The experimental results show that data distribution has a signi cant in uence on the realtime performance and that the choice of commit protocol clearly a ects the magnitude of this in uence. Among the protocols evaluated, the new optimistic commit protocol provides the best performance for a variety of workloads and system con gurations. 1