SliceNBound: Solving closest pairs and distance join queries in apache spark

Abstract

The (K) Closest-Pair(s) Query, KCPQ, consists in finding the (K) closest pair(s) of objects between two spatial datasets. Recently, several systems that enhance Apache Spark with spatial-awareness have been presented, providing a variety of queries for spatial computation, but not the KCPQ. Since queries are of different nature and one processing technique does not fit all cases, we need specialized algorithms for specific queries that exploit the power provided by parallel systems such as Apache Spark. This paper addresses the problem of answering the KCPQ in Apache Spark, by presenting such a specialized, fast algorithm that can easily be imported in any, spatial-oriented or general, Spark-based system. Furthermore, it presents a variant of this algorithm that solves the Distance Join Query. Experiments and comparison to other solutions indicate that our method is fast and efficient. © 2017, Springer International Publishing AG.