View the presentation slides  (420 Kb)

Abstract

The Australian Research Collaboration Service (ARCS) provides national eResearch infrastructure for supporting Australian researchers. The ARCS Systems Services team is responsible for hosting eResearch applications and providing infrastructure for the ARCS National Grid.

There are a wide range of high-performance computing resources available to Australian researchers and all are slightly different. The ARCS National Grid provides a standard interface for efficiently transferring data and securely launching jobs at these resources.

Globus and GridFTP are the primary technologies used. x.509 Certificates are used for secure authentication and individuals can join Virtual Organisations to access common resources. An information service (MDS) is provided that makes it possible to discover the applications and resources available.

Tools have been developed or customized to make the National Grid much easier to use and more accessible to researchers: Grix manages certificate requests and Virtual Organisation membership; Grisu provides a generic job submission interface and allows customised interfaces to be created for particular applications; the Nimrod portal provides an easy way to submit parameter sweep jobs; and the Gridway metascheduler can automatically match jobs to compatible resources based on MDS information.

ARCS will be working closely with the Australian Access Federation to further simplify authentication and access to resources through the use of SLCS certificates.

This talk will give an introduction to the ARCS National Grid and other activities of the ARCS Systems Services team, the technologies and services supported by ARCS, and present examples of applications that are currently using the ARCS National Grid, including bioinformatics, computational chemistry, engineering, geoscience and high-energy physics.

About the speakers

Daniel Cox is the Interim Systems Services Manager for ARCS, based at eResearch SA. He has a background in software development and has been involved with deployment of ARCS infrastructure including grid and authorisation services. Daniel is particularly interested in making these services easier to use for researchers.

Paul Coddington programmed some of the earliest parallel computers while working on his PhD in computational physics at the University of Southampton in the mid-1980s. He subsequently worked on a variety of projects in computational science and applications of high-performance computing and the Web at Caltech and Syracuse University. In 1997 he moved to the University of Adelaide, where he is a Senior Lecturer and leads the Distributed and High-Performance Computing research group, with interests in parallel and distributed computing and applications, and scientific data repositories. He is currently Interim Director of eResearch SA, and was Deputy Director of its predecessor, the South Australian Partnership for Advanced Computing. He is also Projects Manager for the Australian Research Collaboration Service.