Interval assignment for periodic transactions in real-time database systems

The problem of assigning execution intervals for periodic transactions in real-time databases is discussed. The object value evolution rate and the importance of the object are used as two factors for deciding transaction periods. Two different objective functions are defined to reflect different system design goals. Algorithms for optimizing each objective function are presented. The principle behind these algorithms is to allow the transactions which have higher weights to be executed more often. It is assumed that systems use the rate monotonic algorithm for scheduling transactions. Many other scheduling algorithms, like the earliest deadline first algorithm, can also be used. Some examples using the proposed algorithms are given.