In the environment of embedded mobile real-time database management systems, not only the throughput, but also, more importantly, time constraint must to be concerned. Hard real-time transactions must complete before or at their deadline, best effort on the goal of make the firm and soft transactions meet their deadline. Traditional scheduling strategy is based on the throughput entirely, and is unfit for the real-time environment.In the year of 1973, Liu and Layland who are come from jet propulsion laboratory studied the problem of scheduling periodic tasks with hard deadlines and derived two scheduling algorithms which is called rate monotonic(RM) and earliest deadline first(EDF). They are be proved which is the optimal statistic or dynamic scheduling algorithm separately. In the practical real-time environment, transactions could have hard, firm and soft time constraints which is called mixed real-time environment. And mixed scheduling algorithm is yield to fit this kind for environment. Many people have studied in the area, most of them are based on RM and EDF algorithms, and two main scheduling systems was found. One is bandwidth preserving algorithms and another is slack stealing algorithms.Polling server is the simplest mixed scheduling algorithm. The founding strategy of polling server based on the hits algorithm, which is based on the RM algorithm, scheduling hard real-time transactions with RM and services any aperiodic transactions with polling. A real-time scheduling algorithm based on priority table was also included for selecting which aperiodic transaction is the most valuable to be execute.A prototype has been designed and implemented to justify the algorithm and scheduling strategy which is derived in this paper. |