Charting an Open Future for AI Data Storage
Earlier this year, the Open Flash Platform (OFP) initiative was introduced with inaugural members including: Hammerspace, the Linux community, Los Alamos National Laboratory, ScaleFlux, SK Hynix, and Xsight.
This initiative addresses many of the fundamental requirements emerging from the next wave of data storage for AI. The convergence of data creation associated with emerging AI applications coupled with limitations around power availability, hot data centers, and data center space constraints means we need to take a blank-slate approach to building AI data infrastructure.
- Why Open? Flexibility, choice, and removing the bottlenecks introduced by proprietary systems directly benefit data center efficiency.
- Why Now? While existing architectures have more or less allowed us to absorb the massive growth of data that accompanied the “Internet and Cloud Era,” the “AI Era” is set to make that growth look minuscule.
- How? By eliminating the tolls the storage server architecture has placed on data storage, and optimizing a design to reinforce the capabilities of high-density flash, emergent DPU/IPU chips and standard Linux capabilities, like NFS.
The OFP Initiative is delighted to welcome its newest member, Phison, to the project. As we explore optimizing flash storage for the data center, vendors like Phison are essential to turning our collective vision into technical reality.
“Exploring innovative ways to optimize storage for the data center is the only way to keep pace with the requirements of modern applications,” said Sebastien Jean, CTO, Phison. “An open and standards-based approach, like OFP, should deliver an architecture that maximizes customer choice while also driving down power and cooling needs. At Phison, we specialize in blending multi-domain knowledge like storage, AI, compute, and networking, and bring these solutions to the community through initiatives like OFP.”
- Read this blog to learn more about the initiative and why member organizations are advocating for an open, standards-based approach to storage.
- Listen to the “Data Unchained Podcast” interview with Hammerspace CEO David Flynn on “The Infrastructure Revolution for AI Factories”
Don’t miss seeing the Open Flash Platform in action at this year’s OCP Global Summit, happening in San Jose, CA from Oct 13 – 16. Hammerspace will be in the Xsight booth, C70.
Technology Advancements
Since its launch in July, the Open Flash Platform (OFP) initiative has advanced rapidly, with tangible milestones across hardware and software development.
Prototype Demonstration at OCP
At the October Open Compute Project (OCP) conference, Xsight Labs will demonstrate the first OFP prototype in its whisper suite. The system validated the architectural approach of embedding standards-based, Linux-native NFS capabilities directly with flash media, showing how OFP can reduce latency, improve power efficiency, and simplify integration with AI and HPC workloads.
Early Shipments Expected by End of 2025
The development timeline remains on track, with initial shipments of a limited number of OFP-based systems targeted before the end of 2025. These early units will enable partners and customers to begin testing and validating the architecture in production-scale environments.
Linux Kernel Advancements
Key software progress is also underway. On November 12 at 10:00 a.m., the Linux Foundation will host a webinar with Trond Myklebust, NFS Kernel Maintainer, and Molly Presley, Hammerspace Head of Global Marketing. The session will detail kernel-level capabilities that are leveraged by OFP, including enhancements to NFSv4.2 and pNFS, ensuring that OFP integrates seamlessly into existing Linux-based infrastructures.
OFP Presentation at AI Infrastructure Field Day 3
Watch Hammerspace’s Kurt Kuckein’s presentation at AI Infrastructure Field Day (21:40), which highlights the growing challenges of data storage, power consumption, and cooling in modern data centers, especially with the increasing volume of data generated at the edge.
The core idea behind the OFP initiative is to leverage recent advancements in large-capacity flash (QLC), powerful DPUs (Data Processing Units), and Linux kernel enhancements to create a highly dense, low-power storage platform. This platform aims to replace traditional CPU-based storage servers with a modular design, ultimately allowing for exabyte-scale deployments within a single rack.
The proposed architecture consists of sleds containing DPUs, networking, and NVMe storage, fitting into trays that can be modularly deployed. This approach offers significant improvements in density and power efficiency compared to existing solutions. While the initial concept uses U.2 drives, the long-term goal is to leverage an extended E.2 standard for even greater capacity. Hammerspace is leading the initiative, fostering collaboration among industry players, including DPU and SSD partners, and exploring adoption by organizations like the Open Compute Project (OCP).
Industry Coverage of OFP
Several leading publications and industry analysts have covered the launch of the Open Flash Platform, including:
The Bloor Group/Eric Kavanagh:
Rolling Thunder: The Open Flash Platform
“Like thunder rolling louder with each rumble, the OFP movement is gaining industry momentum. It’s not just a spec. It’s backed by vendors aligned around a shared vision. As more tools, drivers, and use cases emerge, OFP will grow from an alternative option to a default standard for high-performance AI data storage.”
NAND Research/Steve McDowell:
Open Flash Platform Initiative Aims to Redefine Flash Storage for AI
“OFP advocates for a ground-up redesign of flash infrastructure, shifting away from the traditional storage server model to a network-attached, open standards-based approach that directly exposes flash to the storage fabric.”
Forbes/Tom Coughlin:
In AI, No Data Is Ever Cold
“…David Flynn, CEO of Hammerspace at the 2025 SNIA SDC…gave a clear and well thought out talk about the Open Flash Platform, OFP, based upon NFS in Linux and a metadata server with data orchestration in a separate control plane from the data, that would dramatically improve system performance by avoiding system controllers and moving data directly from the storage devices to processors.”
Blocks & Files:
Hammerspace Pushes Open Flash Platform to Rethink AI Data Storage
“…the OFP initiative seeks to combine the advancements in the capabilities of offload engines, NAND technology, and Linux capabilities available today into a design targeting an urgent industry need.”