| 摘 要: 随着科学技术的发展，计算机分布式系统在维持数据库的一致性的问题上广泛应用。在商用系统中，通常 在数据中大量的数据需要经常更新，并且现在流行不间断服务，有必要为用户提供在线交易并行一次性更新服务。针对 当前对大量数据更新效率不高的问题，提出了一种分布式系统中大批量数据时序更新方法，通过时序更新的方法避免一 次性更新和在线事务之间的冲突，先在本地交易执行，然后一次提交联合数据库，减少了交易时间的占用，有着更高的 处理效率。实验证明这种在分布式系统中更新数据方法与分批处理方法相比，数据更新执行时间，在每1000次更新执行 时间会减少为原来的1/80，有很高的应用价值。 |
| 关键词: 数据库；分布式系统；批量处理；分布式事务 |
|
中图分类号: TP399
文献标识码: A
|
|
| Distributed System Data Timing Update Method |
|
YU Xingping，LI Hongjian，YU Tengfei，BI Weihong1,2
|
1.( 1.Shandong Vocational College of Science and Technology, Weifang 261053, China;2.
2.Yanshan University, Qinhuangdao 066004, China)
|
| Abstract: With the development of science and technology,distributed systems are extensively applied in maintaining database consistency.In business systems,mass data need to be updated frequently.Since non-stop service is growing in popularity,it is quite necessary to provide online transaction service with once-and-for-all update to users.To deal with the low update efficiency of mass data,the paper proposes a time-sequence update method of mass data in distributed systems,which can effectively avoid the conflict between the once-and-for-all update and the online transactions.The transactions will be firstly conducted locally before submitting the joint database,which reduces the occupation time of transactions and brings higher processing efficiency.Experiments show that,through the method of updating data in distributed systems(compared with the batch processing method),the execution time can be reduced by 1/80 in every 1000 updates. |
| Keywords: database;distributed systems;batch processing;distributed transaction |
