多数据库事务并发调度算法优化技术研究  

Transaction Concurrency Scheduling Algorithm Optimizations Technology Research in Multiple Database

作　　者：杜立佳[1] 董丽丽[1] 何浩[1] 申艳芬[1] 

机构地区：[1]西安建筑科技大学,陕西西安710055

出　　处：《计算机仿真》2011年第2期393-396,共4页Computer Simulation

摘　　要：研究优化调度数据库中事务问题,为保证多数据库中事务提交顺序,提高事务执行及提交的并发度,研究了事务提交图的调度算法TM2,针对算法TM2中事务管理器无法检测到全局事务与局部事务的间接冲突,提出了优化算法TMO,采用ticket算法在全局事务的每个子事务之间增加数据操作,以便在每个成员数据库的全局子事务之间创造直接冲突,并通过添加保存点恢复中止事务代替重做事务,保持了多数据库中事务的可串行化。通过仿真,对比了两种调度算法的性能。得出算法TMO解决了事务的可串行化问题,提高了事务的并发度,保证了事务的一致性。In order to ensure the submitted sequence of muhi - database transaction and enhance the concurrency of transaction implementing and submitting, presented a transaction - based scheduling algorithm of plan submitting of TM2. For GTM in the TM2 algorithm can not detect the indirect conflict between global transaction and local affairs, an optimization algorithm of TMO is proposed, by using ticket algorithm to increase data manipulation between affairs of global transaction services, to create a direct conflict between the sub - transactions of each member of the global database, and by adding a save point to restore the suspension to replace service redo affairs, which maintains the serializability of multi - database transaction. Through simulations, the performances of two scheduling algorithms are compared. TMO algorithm can solve the issue of affairs serializable, and ensure transaction consistency.

关 键 词：多数据库 可串行化 事务 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]