We propose a new algorithm for external grouping with large results. Our approach handles skewed data gracefully and lowers the amount of random IO on disk considerably. Contrary to existing grouping algorithms, our new algorithm does not require the optimizer to employ complicated or error-prone procedures adjusting the parameters prior to query plan execution. We implemented several variants of our algorithm as well as the most commonly used algorithms for grouping and carried out extensive experiments on both synthetic and real data. The results of these experiments reveal the dominance of our approach. In case of heavily skewed data we outperform the other algorithms by a factor of two.
Dieser Eintrag ist Teil der Universitätsbibliographie.
Das Dokument wird vom Publikationsserver der Universitätsbibliothek Mannheim bereitgestellt.
![[img]](https://madoc.bib.uni-mannheim.de/style/images/fileicons/application_pdf.png)
Suche Autoren in