Selectivity Estimation of Inequality Joins in Databases

Selectivity estimation refers to the ability of SQL query optimizer estimate size results a predicate in query. It is main calculation based on which can select least expensive plan execute. While problem has been known since mid-1970s, we were surprised that there are no solutions literature for selectivity inequality joins. By testing four common database systems: Oracle, SQL-Server, PostgreSQL, and MySQL, found open-source systems PostgreSQL MySQL lack this estimation. Oracle SQL-Server make fairly accurate estimations, yet their algorithms secret. This paper, thus, proposes an algorithm join The proposed was implemented sent as patch be included next releases. We compared implementation with above DBMS three different data distributions (uniform, normal, Zipfian) showed our provides extremely estimations (below 0.1% average error), outperforming other by order magnitude.