Title: Parallel out-of-core sorting and fast accesses to disks

Authors: Christophe Cerin, Olivier Cozette, Gil Utard, Hazem Fkaier, Mohamed Jemni

Addresses: LIPN – UMR CNRS 7030, Institut Galilee, Universite Paris-Nord, 99, Avenue Jean-Baptiste Clement, 93430 Villetaneuse, France. ' Universite de Picardie Jules Verne LaRIA, Bat Curi, 5 Rue du Moulin Neuf, F-80000 Amiens, France. ' Universite de Picardie Jules Verne LaRIA, Bat Curi, 5 Rue du Moulin Neuf, F-80000 Amiens, France. ' Ecole Superieure des Sciences et Techniques de Tunis, Departement d'Informatique, 5AV Taha Hussein Montfluery, 1008 Tunis, Tunisie. ' Ecole Superieure des Sciences et Techniques de Tunis, Departement d'Informatique, 5AV Taha Hussein Montfluery, 1008 Tunis, Tunisie

Abstract: The paper addresses two problems. We investigate the problem of parallel external sorting in the context of a form of heterogeneous clusters then we investigate the impact of efficient disk remote accesses on the performance of external sorting. We explore three techniques to show how they can be deployed for clusters with proportional processor performances. We also validate the READ2 library, an efficient implementation of remote SCSI disk accesses. We derive a new parallel sorting algorithm that is adapted to the READ2 interface. The expected gain of using READ2 is compared to the measured gain for one external sorting implementation.

Keywords: out-of-core sorting; parallel sorting; performance evaluation; READ2; sorting algorithms; data distribution; load balancing; I/O bandwidth; disk bandwidth; remote I/O; system area network.

DOI: 10.1504/IJHPCN.2005.008035

International Journal of High Performance Computing and Networking, 2005 Vol.3 No.2/3, pp.188 - 202

Published online: 10 Nov 2005 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article