Logical design of OLAP cubes

Summary form only given. OLAP cubes are multidimensional data structures, used for fast data analysis in business intelligence. Their design typically follows the analysis needs. Just like with relational databases, some designs carry anomalies with them. Our presentation starts with an overview of the life cycle of the data in a database environment and a short review of both theory and practice of operational (relational) database design. As theory and practice differ quite a lot in this area, we continue to discuss the challenges in OLAP design and how OLAP design differs from traditional database design. In this presentation, we discuss two types of OLAP design anomalies. One is related to inefficiency in the form of structural sparsity, where the dependencies within the data imply structural null values in the cube. The other is related to confusing results, where certain many-to-many relationships participate in the analysis in a way, which yields seemingly confusing results in the analysis. Having identified the sources of these problems, we present certain design methodologies to avoid them.