The major commercial data warehouse systems available today are based on record-oriented relational technology optimized for OLTP applications. Several authors have shown that substantial improvements in query performance for OLAP applications can be achieved by systems based on transposed files (column-oriented) technology, since the dominant queries only require grouping and aggregation on a few columns of large amounts of data. This new assumption underlying data warehouse systems means that several aspects of data management and query processing need to be reconsidered. We present some preliminary results of an industrial research project which is being sponsored by the Italian Ministry of Education, University and Research (MIUR) to support the cooperation of universities and industries in prototyping innovative systems. The aim of the project is to implement an SQL-compliant prototype data warehouse system based on a transposed file storage system. The paper will focus on the optimization of star queries with group-by.