Unleashing the Power of "What If": Cloud-Enabled High Performance Computing Workflows in Digital Twins for Scenario Exploration
Jeff Steward, John Furlong, Rachel Stutz, Russell Cox, Jeremy Highley, Houjun Wang, Connor Johnstone, Patrick McBride, John Noto, Ryan Kelly, Ryan Nguyen, Junk Wilson, Matthew Shaxted, M J Long, Alvaro Vidal Torreria, Stefan F. Gary · 2024
Traditional digital twin systems have focused on offering valuable insights into real-world systems through current ("what now") and forecast ("what next") interactive visualizations. The full potential for scenario exploration ("what if") remains challenging, especially when such scenarios require high performance computing (HPC) resources. In the case where such HPC resources are not dedicated or capacity is limited, bursting such compute to the cloud is an invaluable tool despite the many challenges involved. Such challenges include provisioning cloud hardware with the appropriate topology, network fabric, etc.; monitoring and controlling cost; and optimizing performance. This paper proposes a novel approach by integrating cloud-enabled workflows into digital twins using the Parallel Works HPC platform, enabling users to actively test "what if" scenarios to visualize and optimize outcomes. This work was done as part of NOAA’s Earth Observation Digital Twin (EO-DT) prototype effort.