Dr Jonathan Woithe1, Dr David Dossett2, Dr Marcus Ebert3, Prof Paul Jackson1, Prof Martin Sevior2
1University Of Adelaide, Australia, 2University of Melbourne, Australia, 3University of Victoria, Canada
Biography:
Jonathan obtained his PhD in physics at the University of Adelaide, and has worked extensively on the development of Atmospheric radar systems in the years since. He has been an active member in the Australian Open Source community, including involvement with Linux Australia. Most recently he has taken on a role at the University of Adelaide in the High Energy Physics group, overseeing a local compute cluster and the development of the Australian Grid Computing site to support the ATLAS and Belle II projects.
https://orcid.org/0009-0008-2836-1162
Abstract:
Experimental high energy particle physics research has significant compute and storage needs. Compute capacity is required both to process raw collision data from experiments attached to particle accelerators, and to analyse and model those collisions. In the case of the ATLAS experiment attached to the European Large Hadron Collider, this involves processing more than 25 petabytes of data per week. In addition, experiments such as Belle II (attached to the Japanese SuperKEKB accelerator) and ATLAS each require tens of petabytes of storage per year to hold raw data, reconstructions and analysis products for later reference or reuse. The necessary high performance computing infrastructure to support this work is provided by over 150 distributed sites around the world which collectively form the global computing grid.
With cloud-based resources becoming ubiquitous, using such infrastructure to create a grid computing site has become attractive from economic, logistic and technical management perspectives. After outlining the technical requirements of a grid computing site, we will describe the Australian cloud-based implementation employed for the "AU-Melbourne" site which currently provides resources for the ATLAS and Belle II experiments.