Towards a model of concurrency to achieve temporal consistency in long duration transaction processing in real time active database systems

Abstract This paper presents modeling of the transaction in the real time active database management systems. Active database with ECA rule has been found to provide an elegant framework to capture semantics of many real life applications. Real time database management system functions as a repository of data and provides storage and retrieval of data. Transactions, which update the database, are assigned deadlines with suitable slackness. Performance of real time database system mainly depends on how often the transaction misses their deadline. In the applications like situation assessment and tracking, trading and stock control needs actions to be taken in stringent time frame for the full benefit of the system. A long duration transaction takes relatively long time to complete its computation even without interference from other transactions. This makes long duration transaction vulnerable to failures and consequent recovery procedure is also complicated. In this paper, we will discuss how long duration transactions derives their deadlines and how ECA rules of active databases helps in processing long duration transactions.