AARNet’s partnership with supercomputing centres, universities and research institutes allow researchers to connect to some of the fastest supercomputers in Australia and undertake globally competitive research.
Supercomputers use many thousands of central processing unit (CPU) and graphic processing unit (GPU) technologies in tandem, achieving in minutes what a general-purpose computer would need weeks, months or even years to accomplish. HPC aids in computationally difficult tasks like modelling the properties of atoms and complex weather systems, and mapping genomes. This information helps researchers uncover the mysteries of the universe, predict the impacts of climate change, improve our response to natural disasters and track the spread of viruses.
In the era of exascale computing, the need for ultra-fast, scalable, high-capacity network connectivity for moving large-scale research data to and from supercomputing centres for analysis has never been greater. This is where AARNet, Australia’s national research and education network, plays a vital role, as the data connector for the research community. AARNet network services connect research facilities across the country to the computing power of Australia’s Tier-1 supercomputers at the National Computational Infrastructure (NCI) based at the Australian National University in Canberra and to the Pawsey Supercomputing Research Centre in Perth, enabling life-changing research projects.
Work investigating rare diseases by researchers at the Garvan Institute of Medical Research whole-genome sequencing facility relies on powerful technology and network connectivity. The data produced by whole genome sequencing is highly complex and goes through multiple stages of analysis and quality control. Some of this is done at Garvan, while other stages are run at dedicated supercomputing facilities including the NCI in Canberra. AARNet partnered with the University of New South Wales IT team and Garvan to design and deploy the network infrastructure required for Garvan’s human genome sequencing centre, which includes a high-speed AARNet link connecting Garvan to NCI. The link is enabling faster access to data than ever before for researchers aiming to find cures and improve treatments for COVID-19, cancers, and other diseases.
Large-scale data captured by powerful telescopes located at the Murchison Radio-astronomy Observatory in outback Western Australia travels over a dedicated high-speed AARNet link to Pawsey, more than 700 kilometres away in Perth. Supercomputers map the night sky and detect stellar objects, and data can be shared with researchers around the globe over AARNet’s national and international network. To help Australia transition to a lower emissions economy, researchers at Griffith University in Queensland are also using Pawsey and AARNet to explore new materials for enabling more efficient green energy production. The researchers are using one million supercomputer core hours to simulate the atomic action of quantum dots, the world’s most efficient solar cells, and test lead-free alternatives. The team at Griffith connects to Pawsey in Western Australia via AARNet. Pawsey’s ongoing partnership with AARNet provides researchers at universities and research institutes across Australia with high-speed access to Pawsey’s supercomputers and, for the Griffith University team, this was invaluable during Queensland’s COVID-19 lockdowns.
As Australia’s international research impact continues to grow, fast scalable connectivity between supercomputers and research facilities becomes ever more vital. Data volumes are larger than ever and require new techniques to draw value from them. Connectivity for automatic data collection and processing pipelines for major projects needs to be near-real-time and scalable to handle flows of petabytes of data. AARNet’s high-speed network is purpose-built for the task, and exists to provide seamless connectivity between researchers, research facilities, instruments and supercomputers to enable discovery and innovation for Australia.
This story was first published on the AARNet website. View the original story.
For more information please contact our contributor(s):