Selectivity estimation for processing SQL queries containing having clauses

摘要

The estimate of the selectivity of a HAVING clause in an SQL query is carried out within a specified time constraint by determining a selectivity estimate for each member in a sample set of group sizes for the HAVING clause using a probabilistic model based on an assumed value distribution. The selectivity estimates for the groups in the sample set are used to interpolate estimates for all possible group sizes and the estimates are combined based on an assumed known group size distribution to provide an estimation of the selectivity for the HAVING clause. Different selectivity estimating approaches are used for each group size based on available time for estimating and on the estimated time to complete the estimate using different techniques.