data analysis (metamorworks/Shutterstock.com)

INDUSTRY INSIGHT

High-performance computing from anywhere

Typically, telework brings to mind long video conference calls, lost productivity and an increase in cybersecurity vulnerabilities. But for many of the government’s most critical workers, there’s even more to consider.

Designers, engineers and data scientists are working on extremely data-intensive projects, from building future spaceflight capabilities and managing national resources to designing public infrastructure. Those tasks require much more than just an internet connection and a government-issued laptop.

The coronavirus pandemic has shifted how and where federal employees work. A survey by IDC showed 52% of respondents indicated that work from home employment models will likely be permanently changed as a result. Feds performing mission-critical tasks must plan to continue remote operations -- including collaborating on data-intensive projects -- both now and when the next crisis hits.

Currently, federal coronavirus researchers depend on a datasets so massive they can only be culled by using supercomputer-like processing power. If that research were to stop because the scientists couldn’t physically come into the lab, a COVID-19 vaccine could be delayed. Similarly, engineers who depend on high-end computer-aided design and manufacturing tools for building 3D virtual models of buildings, bridges and other infrastructure need virtual access to in-office applications to continue their work.

To ensure all agency workers can maintain productivity, agencies closely examine the unique challenges of each federal employee, including those with complex data processing needs, and consider utilizing GPU-accelerated virtual workstations so scientists can continue their work with massive datasets from almost anywhere. Even in a remote-work situation, these workers need access to fully capable 3D workstations and the sensitive data that must remain in the data center. 

Traditional challenges to telework solutions like choppy videos or slow download times shouldn't be overlooked. For some engineers, remote physical workstations can take hours to download or upload work. GPU-accelerated virtual workstations reduce download times from hours to minutes, tangibly impacting an engineer’s ability to meet agency needs.

How does it work? Put simply, software creates virtual GPUs (vGPUs) that enable every virtual machine to share a physical GPU installed on a server. The software can also allocate multiple GPUs to a single virtual machine to power data-heavy, demanding workloads.

While agencies are beginning to embrace GPU-enabled virtual desktops, the private sector has already seen gains. Another recent IDC survey showed that companies experienced a 49% reduction in total operating costs, and it only took them seven months to pay back the technology investment. Productivity is boosted by virtual workstations with GPUs as well, with users reporting a 94% reduction in downtime.

When Clemson University wanted to run barrier-free high-performance computer infrastructure for its research community, for example, it turned to GPU-enabled virtual machines instead of physical servers to build the computing infrastructure.

One of the major benefits Clemson researchers have realized is improved resource utilization. It’s possible to increase GPU resource utilization by 400% through GPU virtualization and judicially job scheduling. This increased resource utilization not only reduces operational costs for the high-performance computing infrastructure, but it also creates more available resources for researchers to run their applications. Thus, by utilizing vGPUs, Clemson was able to save money while opening up more avenues for HPC researchers.

Virtualization technology isn’t new, but it hasn’t always lived up to the intense processing demands of today’s scientists, engineers and researchers. Data intensive workloads have been prohibitively slow to process and handle, but parallel processors have made it possible to conduct this work remotely.

High-performance computing is a critical part of not only future discovery, but also keeping the country up and running. It makes government more effective and efficient, especially in a time when we don’t know when offices will reopen or for how long. Agencies need creative ways to help process all federal workloads remotely. Virtual desktops with vGPUs are one essential tool to help federal workers stay safe and do their jobs from almost anywhere.

About the Author

Anne Hecht is the senior director of product marketing for GPU virtualization software at NVIDIA.

Featured

  • Russia prying into state, local networks

    A Russian state-sponsored advanced persistent threat actor targeting state, local, territorial and tribal government networks exfiltrated data from at least two victims.

  • Marines on patrol (US Marines)

    Using AVs to tell friend from foe

    The Defense Advanced Research Projects Agency is looking for ways autonomous vehicles can make it easier for commanders to detect and track threats among civilians in complex urban environments without escalating tensions.

Stay Connected