On the mismatch between multidimensionality and SQL

ROLAP tools are intended to ease information analysis and navigation through the whole Data Warehouse. These tools automatically generate a query according to the multidimensional operations performed by the end-user, using the relational database technology to implement multidimensionality and consequently, automatically translating multidimensional operations to SQL. In this paper, we consider this automatic translation process in detail and to do so, we present an exhaustive comparative (both theoretical and practical) between the multidimensional algebra and the relational one. Firstly, we discuss about the necessity of a multidimensional algebra with regard to the relational one and later, we thoroughly study those considerations to be made to guarantee the correctness of a cube-query (an SQL query making multidimensional sense). With this aim, we analyze the multidimensional algebra expressiveness with regard to SQL pointing out the features a query must satisfy to make multidimensional sense and we also focus on those problems that can arise in a cube-query due to SQL intrinsic restrictions. The SQL translation of an isolated operation does not represent a problem, but when mixing up the modifications brought about by a set of operations in a single cube-query, some conflicts derived from SQL could emerge depending on the operations involved. Therefore, if these problems are not detected and treated appropriately, the automatic translation can retrieve unexpected results