一种基于数据质量的多处理器平台实时更新事务调度算法

Quality of data based scheduling for real-time update transactions on multiprocessor platforms

提出一种多处理器平台上基于数据质量的实时更新事务全局调度算法(MU-DA)。数据质量根据时态对象的无效程度来定义。算法通过合理地预分配各事务执行所需处理器资源以及动态控制更新实例的接纳和执行使系统数据质量最大化。研究结果表明:MU-DA算法在各种事务集负载下均能保证较高的数据质量;在高负载设置下,MU-DA算法的系统数据质量与用户事务质量均远比基准算法MU-D与MU-SA的高,能够很好地满足用户事务在数据实时性方面的要求。

A quality of data （QoD） aware algorithm MU-DA was proposed to globally schedule the real-time update transactions on multiprocessors. The QoD measures the degree of invalidity of temporal data objects. To maximize the QoD, the algorithm pre-allocates resources of processors to update transactions, and then judiciously admits and schedules the update instances in the runtime. The results show that the proposed algorithm can guarantee high data quality under different system workloads. In particular, the system＇s QoD and user transactions＇ quality are much better than those of the baseline algorithms （MU-D and MU-SA） under high workloads, thus it can satisfy the data timeliness requirements of user transactions.