Resolving conflicts with committing transactions in distributed real-time databases

Abstract

In a distributed real-time database system, if data conflicts occur between executing and committing transactions, the performance can be severely affected. In this paper, we propose an approach, called deadline driven conflict resolution (DDCR), which integrates concurrency control and commitment management in resolving data conflicts between executing and committing transactions, while at the same time maintaining the schedules to be serializable. With DDCR, many of the data conflicts can be alleviated, and concurrent execution of transactions is allowed to access data items being held by committing transactions. The impact of temporary failures occurred during the commitment of a transaction on other transactions is reduced by reversing the dependencies between the transactions. Simulation experiments have been performed and the results show that DDCR can significantly improve the system performance especially when the duration for the voting phase is long.