Enhancing UNICORE storage management using hadoop distributed file system

  • Authors:
  • Wasim Bari;Ahmed Shiraz Memon;Bernd Schuller

  • Affiliations:
  • Rheinisch-Westfälische Technische Hochschule, Aachen, Germany;Institute of Advanced Simulation, Jülich Supercomputing Centre, Forschungszentrum Jülich GmbH, Jülich, Germany;Institute of Advanced Simulation, Jülich Supercomputing Centre, Forschungszentrum Jülich GmbH, Jülich, Germany

  • Venue:
  • Euro-Par'09 Proceedings of the 2009 international conference on Parallel processing
  • Year:
  • 2009

Quantified Score

Hi-index 0.00

Visualization

Abstract

UNICORE is a state of the art and well tested Grid middleware, designed for seamless and secure access to distributed resources, applications and data, in an easy to use fashion. A wide variety of UNICORE applications for example in bio-informatics generate and compute huge amounts of data. These large amounts of data are not easy to manage reliably and efficiently with the default UNICORE storage system which is using a standard file system. Hence, the current UNICORE does not support a scalable distributed storage system so far. We have integrated Apache Hadoop and its supported distributed storage/ file systems into the UNICORE storage management service. Thus allows to build a UNICORE storage system providing data replication, disaster recovery, durability and elasticity. In this paper we will present the architecture and operation of a prototype called UniHadoop, which provides the integration of UNICORE and the distributed storage systems (DSS) supported by Hadoop and highlight its potential in usage scenarios.