摘要
在类似闭环控制的硬实时数据库应用环境,实时事务具有一定的静态可预报性,其中实时事务的可调度性分析是维护实时数据库时间正确性的基础。通过利用抢占阈值,提出了一种新的实时事务处理模型,它集成了CPU调度和数据调度,实现离线并发控制,具有单阻塞的特征与好的静态可预测性,并有利于降低事务系统的负载和改善可调度性。进一步由此建立了实时事务的静态可调度性分析模型以及求最优可行调度的整数规划模型,该模型有利于达到实时事务调度的整体优化。
In hard real-time database applications for closed loop control environments, real-time transactions may have some predictabilities, and schedulability analysis is critical to maintaining their time correctness. This paper proposes an integrated real-time transaction-processing model by use of preemption threshold. Its special off-line concurrency control can bound priority:inversion time, and may reduce the run-time overhead and improve schedulability. Based on schedulability analysis, this paper also presents the conditions of feasible transaction scheduling for given real-time transaction set, and an integer programming model for constructing optimal scheduling, which is of benefit to the whole optimization of scheduling.
出处
《计算机科学》
CSCD
北大核心
2005年第10期110-113,共4页
Computer Science
基金
国家自然科学基金(60073045)