Improved Histograms for Selectivity Estimation of Range Predicates：（改进的直方图范围谓词的选择性估计）.pdfVIP

Improved Histograms for Selectivity Estimation of Range Predicates Viswanath Poosala Yannis E. Ioannidis University of Wisconsin-Madison University of Wisconsin-Madison poosala@ yannis@ Peter J. Haas Eugene J. Shekita IBM Almaden Research Center IBM Almaden Research Center peterh@ shekita@ Abstract database. The resulting size-estimation errors may under- mine the validity of the optimizer’s decisions or render the Many commercial database systems maintain histograms to sum- marize the contents of relations and permit efficient estimation of user interface application unreliable. Earlier work has sh own query result sizes and access plan costs. Although several types of that errors in query result size estimates may increase expo - histograms have been proposed in the past, there has never be en a nentially with the number of joins [IC91]. This result, in con- systematic study of all histogram aspects, the available choices for junction with the increasing complexity of queries, demon- each aspect, and the impact of such choices on histogram effective- strates the critical importance of accurate estimation. ness. In this paper, we provide a taxonomy of histograms that cap- Several techniques have been proposed in the literature to tures all previously proposed histogram types and indicate s ma