2022 ACM PODS Alberto O. Mendelzon Test-of-Time Award

Abstract

Citation. This paper took research on a fundamental problem in database research join query processing in a new direction. Its motivation was the bound on join query size of Atserias, Grohe, and Marx, now known as the AGM bound (FOCS 2008). This raised the question of whether a join algorithm can achieve a worst-case running time in line with this bound. This paper presents an algorithm that achieves this bound, while showing that traditional query plans cannot achieve it. In the process, they connect join processing questions with geometric inequalities, a connection that has proven fruitful in subsequent work. The algorithmic contribution in this paper almost immediately resonated within database applications when it was observed that a join algorithm recently implemented in industry, Leapfrog Triejoin, achieves a similar optimality guarantee. This led to a line of papers and implementations of join algorithms building off the ideas in the paper. The contribution of the paper to analysis of join queries has arguably been more profound – the connection between join query processing, geometric inequalities, and worst-case size bounds have been subsequently explored in many other contexts, including in the presence of integrity constraints. This work has already been honored with a “Gems of PODS” talk in PODS 2018: the conference paper, journal paper in JACM, and SIGMOD record survey article discussing later developments are all highly cited. This underlines the fact that this paper represented a major departure point for research in database theory.