An interface to grid file method for datacube storage and range sum query

In this work, an interface to the Grid File for datacube storage and Range sum queries for storing multi-dimensional datacubes for Online Analytical Processing (OLAP) is proposed. The grid file has a natural appeal for the storage of datacubes as it confirms to the grid structure of the datacubes. The proposed approach adapts itself to the insertions of new data by splitting the cells dynamically. It minimizes the drawback of the traditional grid files in dealing with uneven data while enjoys its well known advantages of easy implementation and maintenance. The experiment results shows that the proposed method performs much better than those datacubes implemented by R*- trees in file size, construction time, and query response time for evenly and unevenly distributed data.; In this study, we choose to use the grid file to store the multi-dimensional data cube. The grid file conforms to the data cube naturally. It is simple, does not require a complex index structures, and allows direct access to desired blocks. The data space is partitioned by a grid and each grid cell, which contains a number of data, is stored as a disk block. When a disk block is overflowed, the grid cell splits as necessary. The splitting is dynamic and hierarchical. A grid directory is stored in memory of which each element stores the aggregated value of the measure attributes of the corresponding grid cell. Consequently, only when a grid cell is partially overlapped with the range-sum query, need we access the cell on disk. We have performed extensive experiments on the grid file approach. The results show that our grid file structure not only is simple but also have good performance in disk access and CPU time.