Bridging the Gap between OLAP and SQL

In the last ten years, database vendors have invested heavily in order to extend their products with new features for decision support. Examples of functionality that has been added are top N [2], ranking [13, 7], spreadsheet computations [19], grouping sets [14], data cube [9], and moving sums [15] in order to name just a few. Unfortunately, many modern OLAP systems do not use that functionality or replicate a great deal of it in addition to other database-related functionality. In fact, the gap between the functionality provided by an OLAP system and the functionality used from the underlying database systems has widened in the past, rather than narrowed. The reasons for this trend are that SQL as a data definition and query language, the relational model, and the client/server architecture of the current generation of database products have fundamental shortcomings for OLAP. This paper lists these deficiencies and presents the BTell OLAP engine as an example on how to bridge these shortcomings. In addition, we discuss how to extend current DBMS to better support OLAP in the future.