The federal government has flagged possible upgrades to supercomputing, network and big data infrastructure to meet Australia's research priorities over the next decade.
An issues paper [PDF] released yesterday seeks input on the research domains Australia should prioritise.
This will then determine what kinds of research infrastructure needs to be upgraded or replaced to enable Australia to maintain or attain leadership in those fields.
The paper acknowledges high-performance computing (HPC) capabilities as a vital piece of infrastructure necessary for Australian researchers to “maintain a competitive edge globally”.
While Australia’s two tier-1 HPC centres - the National Computational Infrastructure and the Pawsey Supercomputing Centre - have satisfied capacity requirements to date, the paper believes the facilities could synchronise future upgrades “in a planned, alternating fashion”.
This would ensure there is always HPC capacity available to researchers.
The paper also looks at the possibility of creating a “more integrated, coherent and reliable platform to deal with data-intensive, cross-disciplinary and global collaborative research”.
Many research fields it considers as priorities are becoming more data-driven.
A range of national services will be needed to support the future of data-intensive research, including cloud-based collaboration environments, computation and storage, trusted data repositories, virtual laboratory support, and the publishing of research data.
These new services could build on investments in existing research storage services and be run in conjunction with major HPC facilities.
Representatives of research compute organisations including QCIF, Intersect, Pawsey, eRSA and VicNode told iTnews they were considering the paper.
CAUDIT, which is the peak representative body for higher education CIOs and IT directors at higher education and research organisations across Australia, New Zealand and the South Pacific, welcomed the paper and indicated they would also participate in the process.
The issues paper also examines whether research network infrastructure will need to be extended in future.
It raises concerns over access to high-bandwidth resources, as well as the potential for capacity upgrades on intercapital routes and into the Asia Pacific region.
" International connectivity shows a strong bandwidth bias (10 and 40 Gbps) to North America with a relatively narrow (2.5 Gbps) connection to Singapore as Australia’s Asian gateway," the paper noted.
"Given the importance of Asia, and particularly China, as research partners with large data generation capacity, an assessment of projected international data transfers over the next decade would be worthwhile."
The issues paper will be open for comment until 5.00pm (AEST) on September 9.