OLAP Query Routing and Physical Design in a Database Cluster
|Title||OLAP Query Routing and Physical Design in a Database Cluster|
|Author(s)||U. Röhm, K. Böhm, H.-J. Schek|
|Booktitle||In: C. Zaniolo, P.C. Lockemann, M.H. Scholl, T. Grust (Eds.), Advances in Database Technology, Proceedings of the 7th International Conference on Extending Database Technology, EDBT 2000, pages 254-268Konstanz, Germany|
|Organization||Lecture Notes in Computer Science 1777, http://link.springer.de/link/service/series/0558/tocs/t1777.htm, Springer Verlag, Berlin, 2000|
AbstractThis article quantifies the benefit from simple data organization schemes and elementary query routing techniques for the PowerDB engine, a system that coordinates a cluster of databases. We report on evaluations for a specific scenario: the workload contains OLAP queries, OLTP queries, and simple updates, borrowed from the TPC-R benchmark. We investigate affinity of OLAP queries and different routing strategies for such queries. We then compare two simple data placement schemes, namely full replication and a hybrid one combining partial replication with partitioning. We run different experiments with queries only, with updates only, and with queries concurrently to simple updates. It turns out that hybrid is superior to full replication, even without updates. Our overall conclusion is that coordinator-based routing has good scaleup properties for scenarios with complex analysis queries.