Using High Performance Remote Research Desktops to Lower the Barrier of Entry for HPC Resources
Wednesday, June 30, 2021 12:20 PM to 12:40 PM · 20 min. (Africa/Abidjan)
Stream#2
HPC Workflows
Information
Contributors:
Abstract:
Indiana University operates a remote Linux desktop environment called the Research Desktop. This service provides students and researchers access to graphical applications, in a low latency and highly interactive fashion. The system has been designed to lower the barrier of entry and broaden adoption of traditional HPC and high-throughput computing environments. While the service provides all the normal HPC command line tools and allows for direct job submission to the HPC machines, it is designed for users to run computationally expensive applications like Matlab, Comsol Multiphysics, R-Studio and Jupyter right on the desktop. This provides users with an environment that looks familiar to what they know from Microsoft Windows or Mac OSX, while offering the storage and compute resources of HPC compute nodes. The service allows for long run times, as well detaching and re-attaching to a session, making it easy for users to start and monitor long running computational workflows. The talk will provide an architectural overview, use cases and experiences for operating such an environment.
Abstract:
Indiana University operates a remote Linux desktop environment called the Research Desktop. This service provides students and researchers access to graphical applications, in a low latency and highly interactive fashion. The system has been designed to lower the barrier of entry and broaden adoption of traditional HPC and high-throughput computing environments. While the service provides all the normal HPC command line tools and allows for direct job submission to the HPC machines, it is designed for users to run computationally expensive applications like Matlab, Comsol Multiphysics, R-Studio and Jupyter right on the desktop. This provides users with an environment that looks familiar to what they know from Microsoft Windows or Mac OSX, while offering the storage and compute resources of HPC compute nodes. The service allows for long run times, as well detaching and re-attaching to a session, making it easy for users to start and monitor long running computational workflows. The talk will provide an architectural overview, use cases and experiences for operating such an environment.
