High-performance computer commissioned for big data analysis

XENON Systems has won a competitive tender to build a bespoke high-performance computer (HPC) for The University of Queensland (UQ), said to be capable of processing and analysing larger datasets than ever before.

The FlashLite computer was designed by Professor David Abramson, the director of UQ’s Research Computing Centre. XENON will build the computer and preconfigure the operating systems and cluster middleware at its headquarters in Melbourne, before delivering it to UQ in July this year.

Inspired by the US National Science Foundation’s supercomputer, Gordon, FlashLite will include three main innovations: high-speed flash memory (instead of spinning disk); large amounts of high-speed main memory; and software-shared memory. These features will help the computer to deliver breakthroughs in research areas such as cardiac disease, climate change and astrophysics.

“FlashLite will solve a problem that is everywhere these days: big data, and how to exploit it in critical research,” said Professor Abramson.

“We asked whether existing computing systems were fundamentally built the right way to leverage big data. XENON proposed a suitably different solution that offered genuine innovation to help this.”

XENON will deliver a system that includes ScaleMP software to provide huge memory mainframe-style processing benefits in a modern cluster system using next-generation technologies. Compute nodes in FlashLite can be flexibly aggregated together into larger ‘supernodes’ using ScaleMP’s vSMP (versatile symmetric multiprocessing) software. This software combines multiple physically separate servers into one single virtual high-end SMP machine.

FlashLite was funded by the Australian Research Council in conjunction with CSIRO, Griffith University, Monash University, Queensland Cyber Infrastructure Foundation, Queensland University of Technology, UQ and the University of Technology, Sydney. Outside of these stakeholder institutions, a portion of the computer’s capacity will be available to Australian researchers via the National Computational Infrastructure’s National Computational Merit Allocation Scheme (NCMAS).