基于EDF的实时数据库动态容错调度算法

实时数据库系统的事务调度过程中,对于即将完成的事务的抢占会造成CPU时间的浪费,降低系统的性能.针对实时数据库中的周期性实时事务提出了一种PEA(preemptive estimate algorithm)软件容错调度算法,算法基于EDF(earliest deadline first)进行事务调度,并结合负载优化算法进行适当调整,采用抢占评估策略来确定是否允许事务抢占,以最大化系统的资源利用率.通过实验测试,证明其具有良好的性能,能有效提高事务的成功率.

Dynamic Fault-tolerant Scheduling Algorithm Based on EDF in Real-time Database System

In real-time database system,transactions must be scheduled under their deadline.In real-time transaction scheduling,preemption to the nearing completion transaction may lead to the waste of CPU time and reduce the system's performance.For the periodic transactions scheduling in real-time database,a PEA(preemptive estimate algorithm) software fault-tolerant algorithm was proposed.Based on EDF(earliest deadline first) and combined with the load optimized algorithm for adjusting,this algorithm used a preemptive estimate strategy to determine whether preemption was permitted,so as to maximize the utilization of the system.The simulation experiments show that the algorithm has better performance and can enhance the success rate of primary task effectively.