Priority-based speculative locking protocols for distributed real-time database systems.

With globalization, multinational networked organizations' need for exchange of information has led to the emergence of applications that are heavily dependent on globally distributed and constantly changing data. Such applications include, stock trading, Computer Aided Design and Manufacturing (CAD/CAM), online reservation systems, telecommunication systems, e-commerce systems and real time navigation systems. These applications introduce the need for distributed real time database systems (DRTDBS) which must access/manipulate data spread over a network in addition to meeting the real time constraints and maintaining database consistency. In order to improve performance within DRTDBS, attention needs to be given to concurrency control mechanism and transaction's time constraints. A number of protocols have been suggested in recent years to address these issues. One of the proposed protocols, Speculative Locking (SL), has especially demonstrated the capability of improving performance within Distributed Database System by allowing parallelism between conflicting transactions without violating serializability. This research extends SL by giving it the capability of taking a transaction's priority into consideration when scheduling transactions. In addition, a nested transaction model is used to access the data that is distributed across the network. We propose two new Priority-based Speculative Locking protocols: (1) Preemptive Speculative Locking (PSL) and (2) Priority inheritance Speculative Locking (PiSL). PSL extends SL by allowing any incoming higher priority transaction to preempt and abort any lower priority transaction in case of lock conflict thus giving the higher priority transaction a chance to meet the deadline. PiSL, on the other hand, attempts to prevent any wasted work by avoiding preemption by a higher priority transaction. Instead, the lower priority transaction inherits the priority of the blocked transaction. This gives both transactions an opportunity to meet their deadline whenever possible.