一种Shared-Nothing分布式数据库的构建方法

A Construction Method of Shared-Nothing Distributed Database

分布式数据库能够以相对低廉的代价满足海量数据分析处理的性能需求,兼具良好的可扩展性。文中采用Shared-Nothing架构及MySQL数据库来实现分布式数据库访问层的功能架构及模块设计,提出了使用列存储机制提升分布式数据库系统查询性能的思路,包括数据存储的方法及策略。对所设计实现的系统进行了基准性能测试及扩展性能测试,结果表明,文中所实现系统相较传统列式数据库具有出色的查询性能表现,同时具有优秀的可扩展性,能够以较低的代价满足海量数据分析处理所带来的额外数据库性能提升要求。

Distributed database can meet the performance demands of massive data analysis and processing with relatively low cost, with good scalability. It uses Shared-Nothing architecture as the overall system architecture and the MySQL database architecture to achieve module design of distributed database access layer functions. Put forward the idea of using the column storage mechanism to enhance the query performance of distributed database systems, including data storage methods and strategies to achieve them. It conducted the testing for the designed system from the benchmarks and scalability. The performance results show that this system is excellent in query perform- ance and scalability. This system would meet the demand of the additional database performance upgrading requirements brought by mas- sive data analysis with lower cost.