PEAK:AIO has launched PEAK Open pNFS, the first fully open source, vendor-agnostic metadata server for parallel NFS, developed with Los Alamos and Carnegie Mellon.
At SC25, PEAK:AIO announced the launch of PEAK Open pNFS, the industry’s first fully open-source, vendor-agnostic metadata server for parallel NFS (pNFS) — a landmark development that removes one of the last proprietary barriers in scalable AI and HPC storage.
Developed in collaboration with Los Alamos National Laboratory and Carnegie Mellon University, PEAK Open pNFS offers a community-driven, standards-based path to building large-scale, high-performance data systems. It provides organisations with a frictionless, open route from edge to exascale, ensuring no vendor lock-in and advancing the broader mission of open infrastructure in high-performance computing.
The system’s modular, horizontally scalable architecture allows both metadata and storage performance to grow seamlessly with workload demand. By addressing the entire pNFS stack rather than just the client, PEAK:AIO delivers linearly scalable performance and eliminates metadata bottlenecks that limit traditional file systems. Together with the AI Data Server, the solution forms an end-to-end scalable storage platform that keeps pace with modern AI and HPC requirements.
Gary Grider, High Performance Computing Division Leader at Los Alamos National Labs, said: “Open-source solutions like PEAK:AIO’s PEAK Open pNFS promise to address the growing requirements and performance needs and are a very important step towards achieving a flexible base on which community innovation can proceed.”
Roger Cummings, President and CEO of PEAK:AIO, added: “PEAK Open pNFS removes metadata bottlenecks while our AI Data Server ensures storage performance scales alongside it, without compromise.”
