多数据中心的远程数据同步机制研究

发布时间：2018-01-30 01:33

本文关键词： 远程数据同步多数据中心异地多活 SQL捕获一致性连接驱动　出处：《中国民航大学》2017年硕士论文　论文类型：学位论文

【摘要】：随着云计算和大数据的快速发展,异地多活数据中心成为传大型数据中心的发展趋势。与传统的数据灾备中心不同,异地数据中心不仅有容灾能力,还有实时业务分流的作用。因此,在建设新一代民航旅客信息服务系统时,提出了构建异地多活数据中心的需求,而实现异地多活的关键是高效的远程数据同步机制,论文正是基于这种关键需求而开展研究。本文首先调研某互联网公司、某运输公司售票中心等典型多数据中心的运行模式,结合各自的业务特点、容灾能力、一致性级别、每个数据中心承担的角色以及如何与其它数据中心协调工作等方面,提出了同城主备、两地三中心、异地双活、异地多活的多数据中心运行模式。其次,比较了这四种模式的特点以及适用场景,并结合民航旅客服务信息系统的不同业务需求,分别构建了主从同步的异地双活数据中心运行模式和对等同步的异地多活数据中心运行模式。然后针对现有远程数据同步的方法效率低且不能很好的支持异构数据库之间同步的现状,提出了一种新的远程数据同步方法-基于驱动程序SQL捕获的远程数据同步机制,研究了从驱动程序中捕获SQL的方法,针对其中可能出现的数据不一致的情况,提出了相应的一致性校验算法。最后采用线程休眠模拟网络延迟,编写测试应用程序,在同构和异构数据库之间做了大量的同步实验,结果表明,该机制能提高远程数据库同步效率,并支持异构数据库同步,为云计算大数据时代多数据中心建设提供技术支持。
[Abstract]:With the rapid development of cloud computing and big data, multi-live data center becomes the development trend of large-scale data center. Different from the traditional data disaster preparedness center, the remote data center not only has the capability of disaster recovery. Therefore, in the construction of a new generation of civil aviation passenger information service system, the need to build multi-live data center is put forward. The key to achieve multi-activity is the efficient remote data synchronization mechanism. This paper is based on this key requirement to carry out research. Firstly, this paper investigates a certain Internet company. A transport company ticket center and other typical multi-data center operation mode, combined with their business characteristics, disaster tolerance, consistency level. The role of each data center and how to coordinate with other data centers and other aspects, put forward the same city master and reserve, the two places three centers, two places, two live, different places more live multi-data center operation mode. Secondly. The characteristics and applicable scenarios of these four modes are compared, and the different business requirements of civil aviation passenger service information system are combined. The operation mode of the master-slave synchronization data center and the multi-live data center operation mode are constructed respectively. Then, the existing methods of remote data synchronization are inefficient and can not support the heterogeneous number very well. According to the status of synchronization between libraries. A new method of remote data synchronization is proposed, which is based on the remote data synchronization mechanism captured by the driver SQL, and the method of capturing SQL from the driver is studied. Aiming at the possible data inconsistency, a corresponding consistency checking algorithm is proposed. Finally, thread sleep is used to simulate network delay and test application is written. A large number of synchronization experiments have been done between isomorphic and heterogeneous databases. The results show that the mechanism can improve the efficiency of remote database synchronization and support heterogeneous database synchronization. For cloud computing big data era of multi-data center construction to provide technical support.
【学位授予单位】：中国民航大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP311.13;TP308

【参考文献】