行列混合存储数据库系统的研究

摘要

通过研究列存储技术的特点,提出了一种行列混合存储数据库系统的设计方案.该方案在存储层设立独立的行存储引擎和列存储引擎,采用早物化技术在数据读出之后将列表转换成行表,然后以行的形式完成后续处理.因此,该方法既获得了列存储的读优势又复用了行数据库系统的成熟部件,降低了开发的风险和复杂度.基于PostgreSQL的原型开发与测试证明了该方案的可行性和有效性.%By analyzing the technical features of column store, this paper proposed a design of mixed storage DBMS. On the storage level, both row engine and column engine were built. After the column table was read, it was transformed to row table, so that subsequent processing could be done. Therefore, the design gets the advantages of column store as well as reuses the mature parts of row DBMS, reducing the risk and complexity of the development. PostgreSQL-based prototype and testing prove the feasibility and effectiveness of the proposed design.