Most Data Warehouses (DW) are stored in Relational Database Management Systems (RDBMS) using a star-schema model. While this model yields a trade-off between performance and storage requirements, huge data warehouses experiment performance problems. Although parallel shared-nothing architectures improve on this matter by a divide-and-conquer approach, issues related to parallelizing join operations cause limitations on that amount of improvement, since they have implications concerning placement, the need to replicate data and/or on-the-fly repartitioning. In this paper, we show how these limitations can be overcome by replacing the star schema by a universal relation approach for more efficient and scalable parallelization. We evaluate the proposed approach using TPC-H benchmark, to both demonstrate that it provides highly predictable response times and almost optimal speedup. © 2012 Springer-Verlag.
CITATION STYLE
Costa, J. P., Cecílio, J., Martins, P., & Furtado, P. (2012). Overcoming the scalability limitations of parallel star schema data warehouses. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 7439 LNCS, pp. 473–486). https://doi.org/10.1007/978-3-642-33078-0_34
Mendeley helps you to discover research relevant for your work.