基于列式存储的闪存数据库查询优化策略

Query Optimization Strategies of Flash Memory Database Based on Column Storage

随着闪存的性能和容量的提高,闪存数据库的研究受到了广泛的关注。闪存具有随机读快、擦写代价大的特性,如何利用闪存的这个特点,减少查询时写闪存的频率,提高数据库的查询效率是闪存数据库查询研究的重要问题。已有的很多查询优化算法主要是按传统的行式存储方式来进行优化的,有一定的局限性。文中采用列式存储,提出了一种新的连接算法。该算法最小化了中间临时表,使投影数据量大大减少,达到了少写闪存、减小擦除代价的目的,从而提高了查询的效率。通过与传统行式算法的比较实验,证明了该算法的优越性。

With flash memory to improve performance and capacity, flash memory database research has been widespread concern. Flash- memory has the characteristic of random read fast and erased characteristics of a large cost. It＇ s an important issues of flash memory database query that how to use flash feature and reduce the frequency of queries written. Many query optimization algorithms have been the main line according to the traditional way. This has limitations. In this paper, column storage, a new connection algorithm is proposed that minimizes the staging table, so that projection data is greatly reduced. It achieves the purpose of less write flash memory and reducing the cost of erasing, so this algorithm improves the query efficiency. A experiment with the comparison of a traditional algorithm proves the superiority of this algorithm.