A probabilistic analysis of transactions success ratio in real-time databases

Abstract

Nowadays, due to rapidly changing technologies, applications handling more data and providing real-time services are becoming more frequent. Real-time database systems are the most appropriate systems to manage these applications. In this paper, we study statistically the behaviour of real-time transactions under the generalised earliest deadline first scheduling policy (GEDF). GEDF is a new scheduling policy in which a priority is assigned to a transaction according to both its deadline and a parameter which expresses the importance of the transaction in the system. In this paper, we focus our study on the influence of transactions composition. Precisely, we study the influence of transaction distribution on the system performances and on approximation of transactions success ratio behaviour by a probability distribution. To this end, we have developed our RTDBS simulator and we have conducted intensive Monte-Carlo simulations.