An extendible multidimensional array system for MOLAP
Tatsuo Tsuji, Akihiro Hara, Ken Higuchi · 2006
In MOLAP systems, multidimensional arrays are employed to store fact tables dumped from the frontend relational database. On these fact tables, various kinds of statistical computations such as aggregate operations can be performed efficiently by utilizing the fast random accessing capability of arrays. This capability depends on that the size of an employed multidimensional array is fixed in every dimension, so a simple addressing function can be used to access array elements. But, if a new column value emerges after constructing the fact table, the existing fixed size multidimensional array cannot involve the value. In this paper, we provide an extendible multidimensional array system for MOLAP. Such an array can extend its size dynamically along an arbitrary dimension without any relocation of existing data. This property enables incremental aggregate operations without relocating any data dumped at the latest time. Some problems in making this system work as a basis for MOLAP are stated, and their countermeasures are proposed.