Autonomous data synchronization in heterogeneous systems to assure the transaction
Autonomous data synchronization in heterogeneous systems to assure the transaction
复制标题
异构系统数据自主同步，保障交易
DOI：
10.1109/hase.1999.809492
复制
发表时间：
1999
期刊：
影响因子：
--
通讯作者：
K. Mori
中科院分区：
文献类型：
--
作者：
I. Kaji;Yongdong Tan;K. Mori
许多公司被迫快速响应不断变化的用户需求。为了给消费者提供更便捷的服务，公司内部或公司之间的合作已经被要求。因此，异构信息系统也被要求集成在一起。通过利用系统之间的异质性和系统之间的合作，可以在不违反每个系统的特性的情况下从中提取新的值。在异构系统中，如数据生成时间或频率不同，则数据同步本身应由各个系统完成，而不依赖于单个协调器，因为系统配置是不断变化的，使用同步数据组合的应用程序组的成员也在不断变化。本文提出了异构系统中的自主数据同步技术。需要各系统数据同步组合的应用对象可以通过与其他同步信息的比较，局部判断哪种数据组合最适合作为同步组合。因此，所有系统只需在本地采集数据并交换同步信息，即可使用一致性组合作为同步数据。此外，每个系统的同步成本（由同步等待时间（SWT）和向后数据利用率（BDU）衡量）变得均匀。由于此建议的方法不依赖于单个协调器，因此即使系统配置动态更改，或者即使SyncApps的成员更改，它也能正常工作。仿真结果表明了三个系统在直线上互连时的成本公平性。
Many companies have been forced to quickly respond to changing user needs. To provide more convenient services to consumers, the cooperation within a company or across companies have been requested. As a result, the heterogeneous information systems are also requested to be integrated together. By utilizing the heterogeneity among systems and by cooperating between systems, it becomes possible to draw out new values from them without violating each system's characteristic. In the heterogeneous systems, such as the data generation timings or frequencies are different, the data synchronization itself should be done by each system without relying on the single coordinator, because the system configurations are always changing, and members of application group which use synchronous data combination are also changing. In this paper we propose the autonomous data synchronization technique in the heterogeneous systems. The application objects which requires the synchronous combination of the data in each system can locally judge which combination of data is most adequate as the synchronous combination by comparing the synchronous information with others. Thus all systems can use the consistent combination as the synchronous data only by collecting the data locally and exchanging the synchronous information. In addition, the cost of synchronization in each system, which is measured by the synchronization waiting time (SWT) and backward data utilization (BDU) becomes even. As this proposed method is not relying on the single coordinator, it works correctly even if the system configuration is dynamically changed, or even if members of SyncApps are changed. The simulation results show the cost fairness when three systems are interconnected in straight line.