The Virtual Desktop Infrastructure (VDI) at NCI is simplifying the research process for many data reliant projects across the country. Researchers are able to use these new virtual desktops to access all of the data stored at NCI without having to transfer it to their personal computers. The VDI runs on NCI's private cloud Tenjin, part-funded by the National eResearch Collaboration Tools and Resources project (NeCTAR). By accessing the data remotely instead of needing to transfer it first, researchers are able to manipulate and analyse it in real time.
When Dr Leo Lymburner from the Australian Geoscience Data Cube went to the USA earlier this year to present his data to the United States Geological Survey, he could log in to the VDI and show them the data in the system. He says "We could do a live demo of our current Landsat satellite data holdings from the other side of the planet, which was pretty cool. They were able to see me log on through NCI and show them a live status of what our Landsat collections look like."
After logging in, researchers can use a standard desktop interface to interact with their data, instead of having to go through a layer of coding first. The interface provides them with many different software tools built in, reducing the need to install software on their local computer. It also provides a number of ways for users to create and share their own virtual environments and specialist packages. This opens up complex computational data analysis to a wide range of researchers lacking the programming skills to otherwise do so.
For Dr Andy Hogg from The Australian National University and the ARC Centre of Excellence for Climate System Science, the big benefit in using a VDI is that he doesn't need to submit a compute job on NCI's main supercomputer, Raijin. Jobs computed on Raijin sit in a queue before being sent off for processing, which is fine for large jobs taking several hours and many computer cores, but is less ideal for interactive data analysis. He says "The climate and ocean models we run on Raijin produce huge amounts of data, which we then need to analyse. That's where the VDIs come in. The large memory capacity of the VDI nodes means that I can customise my analysis and try it out on my data in real time."
For users with only a small amount of technical knowledge, the VDI provides a powerful working environment. It looks and feels like a standard computer desktop, but also provides incredibly performant software tools and hugely varied datasets, upgrading its functionality beyond any regular desktop. By eliminating the time and effort required for data transfers and local computation for each user, the VDIs bring the data analysis work of the scientists to the foreground. For this reason, Dr Lymburner says, "I'm absolutely thrilled with it; I use it on a daily basis. It's changed the way I do my work."