联机分析多维存储结构的研究

RESEARCH ON THE MULTIDIMENSIONAL STORAGE STRUCTURE OF ON-LINE ANALYTICAL PROCESSING

联机分析使用多维数组作为存储结构以加快查询响应时间。为了等同的对待每个维,适应稀疏数据,必须对多维数组进行划分。目前,有两种划分方法。本文分析了它们的优缺点,给出了一种统一的存储结构,实验结果表明,为了达到转换时间短和压缩比高的目的,要选择合适的划分向量和数据块体积。

On-line analytical processing uses multidimensional array as storage structure to speed up response time. To treat every dimension equally and adapt to sparse data, it is necessary to partition multidimensional array. There are two types of partition methods. This paper analyzes merit and demerit of these methods and gives a unified storage structure. The results of experiment show it is important to select appropriate partition vector and size of a data block to obtain high compress ratio and to make translation from text file to multidimensional array rapidly.