Case Study: Clemson University achieves faster emergency evacuation planning with Google for Education

Clemson researchers deploy 2.1 million virtual CPUs on Google Cloud to improve emergency planning

Clemson University worked with Google for Education to tackle a major emergency-planning challenge: how to analyze massive amounts of traffic video quickly enough to support evacuation routing and public safety decisions during disasters. The university team needed scalable high-performance computing resources that could handle large, on-demand workloads far beyond what traditional systems could easily provide.

Using Google Cloud, along with tools like PAW, CloudyCluster, and Slurm, Clemson distributed its traffic-analysis workload across up to 2.14 million virtual CPUs and 133,573 concurrent VM instances. The team processed 210 TB of video in about eight hours, showing that Google for Education’s cloud-based HPC approach could scale rapidly, run cost-effectively with preemptible VMs, and support real-time evacuation planning at a lower cost than many on-premises HPC systems.

Clemson University

Amy Apon

Professor