Research On The Optimizing Methods Of Data Cube

Data cube is the kernel conception of data warehouse and on-line analytical processing (OLAP). It usually needs to pre-compute and save the data cube in disk for speeding up the queries. But the size of data cube will increase explosively when they are all pre-computed and saved. To decrease disk cost and improve queries performance are very important but contradictive goals of data cube research. For the sake of resolving these problems, it needs to explore the effective data cube structures.A new data cube structure named Iso-Basictable Cube(IBC) is proposed here, which is compressed based on semantic relations of cube. It groups the cells of cube into a class named Iso-Basictable Class which aggregate from the same tuple subset of basic table and have the same aggregate value. So it only needs to save one cell of each class into disk and reduces the size of the cube remarkably. Meanwhile, the left cells in IBC preserve the semantic relations by virtue of the compressing mechanism. Organizing the cells of IBC using a tree structure named IBC Cuboid Tree decreases the size farther and speeds up the point query.The usual mode of procedure to dealing with a range query of current data cube structures is that firstly decomposes a range query to a lot of point queries and then performs them one by one and gather these results lastly. The number of point queries decomposed from a range query is the product of all points number included in every dimension range. So the efficiency of range query is not high. Here another data cube structure based on sets operation is proposed, which improving range query performance by means of index and intersection and union operations of sets. It firstly unites the index sets of all dimension values of every dimension query range and then looks for the intersection of sets of ever dimension, so that improves the performance of range query and preserves low disk cost and comparatively fast point queries.