Multiclass transaction scheduling and overload management in firm real-time database systems

Real-Time Database Systems (RTDBSs), have attracted considerable amount of research attention in the recent past and a number of important applications have been identified for such systems, such as telecommunications network management, automated air traffic control, automated financial trading, process control and military command and control systems. Due to the rapidity of change of the system state in such applications, as well as the inherent complexities in controlling such systems (which result in frequent violation of consistency requirements and consequent repeated firings of control actions), it is likely that the transaction load in these systems would be fairly high. Thus RTDBSs need to be equipped with overload management mechanisms. Unfortunately overload management has been a fairly neglected area in real-time systems research in general and real-time database research in particular. In this paper we introduce Adaptive Access Parameter (AAP), a scheduling mechanism for multiclass transactions in real-time database systems that employs an explicit admission control policy to manage overload as well as system bias towards particular transaction classes. We show the theoretical underpinnings behind AAP and then report a thorough performance study that demonstrates AAP's substantial superiority over current algorithms with regard to performance metrics as well as computational overhead.