By offering improved access to high performance computing (HPC) resources, IU is diversifying the range of research projects that use these resources. Robert Henschel, director of Research Software and Solutions in IU’s Research Technologies division calls Research Desktop, commonly known as RED, a solution for researchers looking to enhance their work with the power of Carbonate supercomputing clusters, but with the convenience of a desktop interface.
IU Research Technologies aims to amplify the interdisciplinary capabilities of supercomputing through RED’s user-friendly interface. “In order to get users from a broader science background to use the machine, we need to give them an easy way to use it and this is where RED comes in,” said Henschel. “By not charging for it and by not requiring a project proposal or an allocation process, we’re attracting more people who just want to try something out,” he explained.For example, RED has been used by the IU Physics department to teach undergraduate students how to code, using the “Geany” graphical editor.
Before RED, Henschel says IU’s HPC resources favored scientists with coding skills. “More energy in supercomputing goes into making them efficient, making them scale, making them cost effective, than making them user-friendly,” observed Henschel. RED addresses two primary barriers for those without backgrounds in computer science: a graphical user interface (GUI) and web access. RT also provides high performance storage options, which are particularly useful for complex files. This means, “If you are a graduate student or a faculty member that gets an unusually large dataset that can no longer work with on their laptop, you can access this information online instead,” Henschel said.
RED is also a gateway for scientists and researchers to interact with the batch-computing process. Henschel notes that those who are new to supercomputing often find this process confusing. “For users who are new, who have only ever used computers interactively – by typing something and seeing it appear – the concept of having to specify, submit it, and then wait for it is hard to accept,” said Henschel. By prioritizing user-friendly design, Henschel says response rates are positive, which he hopes sparks a long-lasting research relationship with all of IU’s supercomputing resources.
When I can get a research team excited about using one of our HPC machines because they understand they can solve an order of magnitude size larger problems, that’s an awesome day. Because then they will go beyond RED.
At present, the team is working to move RED off Karst, IU’s high-throughput computing cluster, and is working on moving RED to the Jetstream cloud. Jeremy Fischer of Jetstream reports that the proof of concept is up and running; the next step is HIPAA alignment. Running RED on Jetstream would pair leading-edge cloud technology with more traditional ways of doing research computing. Moving RED to Jetstream offers several advantages, including the ability to stay live during maintenance, to scale up and down due to demand, and to balance loads (pairing heavy users and light users on the underlying hardware).