分布式海量数据管理系统Hypertable底层存储结构分析

Analysis of Lower Store Structure in Hypertable System for Massive Data Distributed Management

通过分析Hypertable的源代码,描述了CellStore存储结构,介绍其读写流程,总结了该结构存在的缺陷,并提出了优化思路。优化步骤主要包括:将关键字数据进行合并,建立关键字到数据值的索引结构;并设置一个关键字快表,用于存储访问频率较高的关键字数据。结果表明:该优化结构可减少冗余数据,提高数据的读写效率。

By analyzing the source code of Hypertable, the Cellstore＇s storage structures have been described. The procedures of reading and writing operations have been introduced. The disadvantages of aforementioned structure have been identified; and the development procedures have been recommended. The development procedures include： combining phase information; building up a phase information table, in order to store high-frequency data. The result shows that the optimal structure can reduce overloaded data in order to improve the efficiency and performance of data reading and writing operations.