| | |
Summary: CORDS: Automatic Generation of
Correlation Statistics in DB2
Ihab F. Ilyas1
Volker Markl2
Peter J. Haas2
Paul G. Brown2
Ashraf Aboulnaga2
1
University of Waterloo 2
IBM Almaden Research Center
200 University Avenue 650 Harry Road, K55/B1
Waterloo, Ontario N2L 3G1, Canada San Jose, CA, 95139
ilyas@uwaterloo.ca marklv,phaas,pbrown1,aashraf@us.ibm.com
Abstract
When query optimizers erroneously assume that
database columns are statistically independent,
they can underestimate the selectivities of con-
junctive predicates by orders of magnitude. Such
underestimation often leads to drastically subop-
timal query execution plans. We demonstrate
|
