Parallel Temporal Joins

Author: Thomas Zurek
Date: March 1997
Published in: "Datenbanksysteme in Büro, Technik und Wissenschaft", Proc. of the BTW'97 (German Database) Conference, Ulm, Germany, GI Informatik aktuell
Publisher: Springer Verlag
Pages: 269 - 278

Download

Abstract

We present a framework for parallel temporal joins. The temporal join is a key operator for temporal processing. Efficient implementations are required in order to make temporal database features attractive and applicable for the many applications that are amenable.

We focus on the temporal intersection as the supertype of temporal joins. In contrast to traditional equi-joins, parallel temporal join processing suffers from tuples being replicated between data fragments. This causes a significant overhead.

A basic parallel temporal join strategy -- derived from traditional approaches -- is refined by two optimisations. The quantitative impacts of the optimisations are evaluated. It is shown that both optimisations together decrease the basic costs significantly.

Keywords

temporal join, parallel join, temporal databases, parallel databases, data partitioning

Thomas Zurek, <tz@dcs.ed.ac.uk>