Real-Time Computation of Advanced Rules in OLAP Databases
In Online Analytical Processing (OLAP) users view data through a multidimensional model known as the data cube, allowing the aggregation of information along different attributes and operations such as slicing and dicing. In-memory OLAP systems keep all relevant data in main memory and also support efficient updates of cube data, enabling interactive planning, forecasting, and what-if analysis. Since usually only the base data is stored and all aggregations and other calculations are computed on the fly, complex computations may seriously downgrade performance. We present an approach that uses graphics processing units (GPUs) as parallel coprocessors for high performance in-memory OLAP operations. In particular, our method accelerates the calculation of compute-intensive rules, which represent business dependencies that are more complex than mere aggregates. In addition to the data structures and algorithms, we describe how to extend the approach to multi-GPU systems in order to scale it to larger data sets.
KeywordsData Cube Fact Table Matching Fact Rule Operation Advance Rule
Unable to display preview. Download preview PDF.
- 2.Bakkum, P., Skadron, K.: Accelerating SQL database operations on a GPU with CUDA. In: Proceedings of GPGPU 2010, pp. 94–103. ACM Press, New York (2010)Google Scholar
- 3.Chaudhuri, S., Dayal, U.: Data warehousing and OLAP for decision support. In: Proceedings of SIGMOD 1997, Tucson, AZ. ACM Press, New York (1997)Google Scholar
- 4.CUDA website, http://www.nvidia.com/object/cuda_home_new.html
- 6.Fernando, R.: GPU Gems: Programming Techniques, Tips and Tricks for Real-Time Graphics. Pearson Higher Education, London (2004)Google Scholar
- 7.Govindaraju, N.K., Lloyd, B., Wang, W., Lin, M.C., Manocha, D.: Fast computation of database operations using graphics processors. In: Proceedings of SIGMOD 2004, Paris, France, pp. 206–217. ACM Press, New York (June 2004)Google Scholar
- 9.Harris, M., Sengupta, S., Owens, J.D.: Parallel Prefix Sum (Scan) with CUDA. In: Nguyen, H. (ed.) GPU Gems 3, pp. 851–876. Addison Wesley, Reading (August 2007)Google Scholar
- 11.Horn, D.: Stream reduction operations for GPGPU applications. In: Pharr, M. (ed.) GPU Gems 2, pp. 573–589. Addison Wesley, Reading (March 2005)Google Scholar
- 12.IBM Cognos TM1, http://www.ibm.com/software/data/cognos/products/tm1/
- 13.Infor PM10, http://www.infor.com/solutions/pm/pm10/
- 14.Jedox Palo Suite, http://www.palo.net
- 16.Lauer, T., Datta, A., Khadikov, Z., Anselm, C.: Exploring graphics processing units as parallel coprocessors for online aggregation. In: Proceedings of DOLAP 2010, Toronto, Canada. ACM Press, New York (October 2010)Google Scholar
- 17.Satish, N., Harris, M., Garland, M.: Designing efficient sorting algorithms for manycore GPUs. In: Proceedings of the IEEE International Symposium on Parallel & Distributed Processing, pp. 1–10 (May 2009)Google Scholar