NetApp has completed its acquisition of PEAK:AIO to support AI clouds and GPU-as-a-service providers handling enormous volumes of data.

The deal adds PEAK:AIO’s metadata architecture and parallel file system expertise to NetApp’s ONTAP platform and its planned NetApp Novus storage system.

AI systems need a constant, uninterrupted flow of data to keep expensive GPUs busy. PEAK:AIO’s technology will help NetApp build storage capable of supporting zettabyte-scale file systems, high concurrency and massive growth in file numbers without creating bottlenecks.

The acquired technology separates metadata from data to allow performance to scale as more users and workloads access storage at once. This will use open standards rather than proprietary clients, while retaining enterprise-grade security, resilience and operational simplicity.

“By combining PEAK:AIO’s expertise in metadata services and parallel namespace innovation with the proven scale, data management capabilities, and enterprise-grade services of the NetApp Platform, we are leapfrogging the market as we deliver our next-generation, zettabyte-scale architecture to support the growth of AI clouds and GPU-as-a-service providers,” said George Kurian (top), Chief Executive Officer of NetApp.

PEAK:AIO builds software-defined storage designed to run on standard hardware and scale from a single server to exabytes. Its technology has been deployed by organisations such as Los Alamos National Laboratory, Carnegie Mellon University, NHS AIDE programme, and Oxford Robotics Institute.

“NetApp and PEAK:AIO were built around the shared conviction that the AI era needs data infrastructure designed to scale to extraordinary levels.”

“With the acquisition by NetApp, we can bring our innovation together with the reach, expertise, and trusted customer relationships of a global data infrastructure leader,” said Mark Klarzynski, Chief Technology Officer and Founder of PEAK:AIO.

NetApp plans to integrate PEAK:AIO’s metadata services and parallel namespace technology into its ONTAP-based architecture, while giving existing ONTAP customers an upgrade path. It expects this to support new capabilities, including NetApp Novus, for operating shared storage in increasingly large AI environments.