基于NDB引擎的MySQL Cluster的部署规则及测试

发布时间：2018-05-13 03:32

本文选题：MySQL数据库 + Cluster　；参考：《山东大学》2015年硕士论文

【摘要】：随着计算机技术和网络的发展,数据库集群技术的应用越来越普遍。这是因为集群在高可用性、高吞吐率、数据一致性和容错性等方面的表现突出。数据库作为企业应用系统中的核心,有着举足轻重的作用。因此我们根据企业的生产需求对高性能的数据库集群系统进行研究是十分必要的。构建数据库集群能够大大提高数据库的吞吐率,还能有效减少数据库请求的响应时间。避免网络服务中高数据访问时产生的单点瓶颈问题。在数据库集群系统的构建与应用中,负载均衡技术是决定集群性能优劣的关键因素,好的负载均衡方案可以高效地将负载分发至集群的各个节点,能大大降低集群因节点故障造成的影响。如今,数据库集群技术在互联网上已经有了较为广泛的应用。现在也有一些应用价值很高的产品和开源软件供我们选择使用。但是由于国外公司的技术封锁,企业购买和维护设备的成本都比较高,而且国内在集群技术的研究和应用方面还处在起步阶段。我们选用MySQL数据库来搭建集群系统,主要是考虑到MySQL数据库以其开源免费的特点受到企业的欢迎。搭建MySQL数据库集群的重难点在于解决系统的负载均衡、数据同步和故障切换上,这些是实现高性能和高可用数据库集群面临的关键问题。本文根据山东省信息化人社建设的规划方案和数据中心总体建设方案的整体需求,通过分析当前社保系统使用的数据库集群技术的特点,结合几种商业数据库集群系统的架构设计,针对用户对社保的具体应用需求,我们设计并实现了一种采用非共享存储架构的基于NDB引擎的MySQL数据库集群系统。该系统是可以构建在廉价设备之上的具有高可用性、良好扩展性和高性能的数据库集群。它的主要特点是支持多客户端的并发访问,向客户端应用提供一个统一的虚拟IP地址,来连接MySQL集群中分布式部署的多个MySQL服务器；通过负载均衡技术实现了系统的高可用性和扩展性,并将集群节点故障的影响大大减小,同时集群的负载调度节点采取主从互备的工作模式,可以保证系统的可持续应用；在数据的更新上,每个节点上的数据会做到同步更新,从而保证了各个数据库中数据的一致性。目前,系统已经部署和测试完成,服务性能稳定,初步达到预期效果。可以有效的进行故障切换,负载均衡和数据同步,但是系统的稳定性还有待提高。如果想在社保系统中使用我们还需要对系统进一步的优化处理。
[Abstract]:With the development of computer technology and network, the application of database cluster technology is becoming more and more common. This is due to cluster performance in high availability, high throughput, data consistency and fault tolerance. As the core of enterprise application system, database plays an important role. Therefore, it is necessary for us to study the high performance database cluster system according to the production demand of the enterprise. Constructing database cluster can greatly improve the throughput of database and reduce the response time of database request. To avoid the single point bottleneck problem caused by high data access in network service. In the construction and application of database cluster system, load balancing technology is the key factor to determine the performance of the cluster. A good load balancing scheme can distribute the load to each node of the cluster efficiently. It can greatly reduce the impact caused by node failure in the cluster. Nowadays, database cluster technology has been widely used on the Internet. There are also high-value products and open-source software that we can choose to use. However, due to the technology blockade of foreign companies, the cost of purchasing and maintaining equipment is relatively high, and the research and application of cluster technology in China is still in its infancy. We choose MySQL database to build cluster system, mainly considering that MySQL database is welcomed by enterprises because of its open source and free. The most difficult problem in building MySQL database cluster is to solve the problem of load balancing, data synchronization and failover. These are the key problems to realize high performance and high availability database cluster. According to the overall demand of the planning scheme of the construction of the information society of Shandong Province and the overall construction plan of the data center, this paper analyzes the characteristics of the database cluster technology used in the current social security system. Combined with the architecture design of several kinds of commercial database cluster system, we design and implement a kind of MySQL database cluster system based on NDB engine, which is based on the non-shared storage architecture, according to the specific application demand of users to social security. The system is a database cluster with high availability, good scalability and high performance. Its main feature is to support multi-client concurrent access, to provide a unified virtual IP address to client applications, to connect the distributed deployment of multiple MySQL servers in the MySQL cluster; The high availability and expansibility of the system are realized by load balancing technology, and the influence of the cluster node failure is greatly reduced. At the same time, the load scheduling node of the cluster adopts the mode of master-slave mutual backup, which can ensure the sustainable application of the system. In the updating of data, the data on each node will be updated synchronously, thus ensuring the consistency of the data in each database. At present, the system has been deployed and tested, the service performance is stable, and initially achieve the desired results. Fault switching, load balancing and data synchronization can be performed effectively, but the stability of the system needs to be improved. If we want to use the social security system, we need to further optimize the system.
【学位授予单位】：山东大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TP311.13

【相似文献】