We propose an index structure, called measure attribute (MA) index, and a query processing technique to improve OLAP query performance. OLAP queries are extremely complicated due to representing the intricate business logic of the company on a huge quantity of data. This is why the efficient query evaluation becomes a critical issue in OLAP systems. Proposed query processing technique supports an efficient evaluation of the star joins and grouping operators known as the most frequently used but very expensive operators in OLAP queries. The MA index is a variation of the path index in object databases and supports index-only processing for the star joins and grouping operators. Index-only processing is a well known efficient technique in the query evaluation area. We implemented the MA index on top of an object-relational DBMS. Performance analysis shows that the MA index provides speedups of orders of magnitude for typical OLAP queries.
展开▼