A method for skew resistant join size estimation

摘要

A method of estimating the query size of two databases T and R is disclosed. The method uses a threshold value to categorize the databases as dense or sparse. A dense-dense procedure is then applied to the two databases to produce a dense-dense estimate (Ad). A sparse-any procedure that suppresses the dense data items coming from database T is performed which produces a first sparse-any estimate (As1). A second sparse-any estimate (As2) is then produced by suppressing the dense data items from database R . Ultimately a query size estimate is produced by combining the dense-dense estimate, the first sparse-any estimate and the second sparse-any estimate.