Research Computing: What is it & how can I use it at UniMelb?

by Bernard Meade

Research computing is about using computational resources to produce better and/or faster research outcomes.  Despite the fact that desktop and laptop computers are still managing to adhere to Moore’s Law, that is, doubling in power roughly every two years, the demand of research computing is growing even faster.  Many researchers and research students, particularly those in STEM departments, are finding that the processing power of their local computer is no longer up to the task of dealing with big data or large, complex simulations.

Research Platform Services can help.

Our Compute platforms include a High-Performance Compute (HPC) cluster and the Melbourne Node of the NeCTAR Research Cloud.  These platforms provide researchers with far more power than the typical desktop computer.

The HPC system, called Edward, allows computationally intensive tasks to be distributed over many physical machines all working together.  Compute jobs might need many cpus, or additional memory, or high-speed access to data for processing.  Researchers can use open-source or commercial software designed to take advantage of this parallel processing, or can develop their own code to optimise performance.  The system is used by many researchers and automatically provides fair and efficient access to everyone.

image

Cloud computing provides highly flexible resource allocation, which is increasingly valuable to researchers and university departments.  Rather than purchasing a powerful computer that will only be used occasionally, a Virtual Machine (VM) can be created whenever it is needed, and deleted as soon as it is not.  Alternatively, a VM might be used for an “always on” service such as a web server.  Housed in an Enterprise-grade data centre, these VMs are every bit as robust and usable as a physical machine, but have many additional benefits.  For example, a VM can be cloned, with many new versions created when needed.  They can also be backed up and restored at a later date, or migrated in real-time to another physical host should the existing server become unstable.

Cloud provides the power you need, when you need, and where you need.  

An example provided by Professor Richard Sinnott, Director, Melbourne eResearch Group:

“As a researcher/research infrastructure provider and educator I am heavily reliant on and indeed greatly benefit from the NeCTAR Research Cloud. With regards to research, I am involved in a broad portfolio of projects covering the full gamut of research endeavours where big data and big compute challenges have to be overcome. NeCTAR has provided a unique underpinning environment for such activities. It is important to emphasise that this is far beyond proof of concept activities. For example, the ENSAT-CANCER project (running and hosted on NeCTAR) has over 66 major cancer centres globally and formed the basis for the recently funded $12m Horizon2020 ENSAT-HT project. With regards to education, I have taught the Cluster and Cloud Computing course at the University of Melbourne since 2013. This course is heavily based on the NeCTAR Research Cloud with specific programming exercises on big data processing of social media (Twitter) data. The student numbers are growing drastically from 51 (2013), 71 (2014) and currently 139 (2015). Without the scalability offered through the NeCTAR Research Cloud such courses would be majorly impacted. To train the next generation of computer scientists it is essential that such national infrastructures exist and importantly are sustained.”