The National Research Platform (NRP) functions as a community-operated, Kubernetes-based fabric known as Nautilus. By aggregating CPU and GPU resources from over 70 participating organizations, the platform allows researchers to access high-performance computing capabilities that often exceed their local institutional limits. Historically, however, the diversity of hardware and networking configurations has created significant integration hurdles for those attempting to contribute new nodes to the cluster.
In section Releases
Internet2 Deploys Standardized AI Infrastructure at U.S. Universities
The University of Nebraska–Lincoln and Sacramento State are integrating new Cisco AI PODs into the National Research Platform, a distributed computing fabric. This deployment introduces a modular, pre-validated hardware model designed to simplify how academic institutions contribute GPU capacity to shared, multi-region research projects across the country.
Cisco AI PODs aim to resolve these technical bottlenecks by providing a standardized, full-stack infrastructure. Each unit arrives pre-validated, combining Cisco UCS servers and high-bandwidth Nexus networking with centralized management via Cisco Intersight. By moving away from custom, hand-assembled nodes toward this repeatable model, Internet2 expects to reduce the engineering overhead required for universities to join the fabric. The current deployment at Nebraska and Sacramento State includes eight NVIDIA H200 GPUs, 128 CPU cores, and 90 terabytes of NVMe flash storage, all interconnected via NVIDIA BlueField-3 DPUs to ensure high-speed data flow across national research networks.
Comments (0)
No comments yet. Be the first!