Hammerspace Introduces an All-New Storage Architecture | Read the Press Release

Hammerspace Newsletter March 2024

Welcome to the March 2024 edition of the Hammerspace Newsletter! 

This month is all about Hyperscale NAS, a new and fundamentally different NAS architecture to power GPU computing at scale and move AI innovation forward. Hyperscale NAS combines the performance and scale of HPC parallel file systems with enterprise NAS standards and features. 

Whereas previous NAS architectures were about scaling capacity, Hyperscale NAS is really about scaling performance and capacity – to feed GPU clusters, CPU clusters, and other use cases that require high speed parallel processing of data at scale.

Looking for a quick overview? Here’s a 3-minute video to learn more. 

Want to dive in for more detail?

  • Read this blog post from Meta Engineering about its GenAI infrastructure, including its use of Hammerspace: “Among other benefits, Hammerspace enables engineers to perform interactive debugging for jobs using thousands of GPUs as code changes are immediately accessible to all nodes within the environment. When paired together, the combination of our Tectonic distributed storage solution and Hammerspace enable fast iteration velocity without compromising on scale.”     
  • Watch the replay of Hammerspace’s session at AI Field Day, where Floyd Christofferson and Chad Smith from Hammerspace talk about how to achieve HPC-class performance to feed GPU-based AI pipelines while leveraging data in place on existing storage resources.
  • Read this contributed article in HPCwire, “pNFS Provides Performance and New Possibilities” to learn about a massive leap forward in advancing data path performance from the Linux community that changes the game globally. 
  • Learn what the top tech and business press publications have to say about Hyperscale NAS, including Blocks & Files, CRN, TechTarget, and Forbes.

Meta engineering has released details on how they designed infrastructure, including their use of Hammerspace, for their LLM training for Llama 2 and Llama 3 as well as AI research and development across GenAI.

Check out this Guest Blog Post from Meta’s Kevin Lee, Adi Gangidi, and Mathew Oldham on their open source contributions and all the components in their infrastructure strategy.

Read more at Blocks & Files: “Meta hooks up with Hammerspace for advanced AI infrastructure project” and Silicon Angle: “Meta unveils two new 24K GPU clusters it’s using to develop more advanced generative AI.

Check out our press release unveiling Hyperscale NAS, the fastest file system in the world for training enterprise AI models at scale. This new category of storage architecture removes storage bottlenecks and provides the high-performance data pipeline needed for GPU computing.

The Hyperscale NAS architecture is in production with systems built on approximately 1,000 storage nodes, feeding up to 30,000 GPUs at an aggregate performance of 80 Terabits/sec over standard ethernet and TCP/IP.  

Read our two-part blog series, “The Need for a New Storage Architecture: Hyperscale NAS” which explains the architecture and what makes it so unique and the advantage of Hyperscale NAS over traditional scale-out NAS architectures.

Hammerspace Hyperscale NAS is now available with NVIDIA GPUDirect storage support. What does this mean? The added support allows enterprise and public sector organizations to leverage Hammerspace software to unify unstructured data and accelerate data pipelines with NVIDIA’s Magnum IOTM GPUDirect® family of technologies, which enable a direct path for data exchange between NVIDIA GPUs and third-party storage for faster data access. By deploying Hammerspace in front of existing storage systems, any storage system can now be presented as GPUDirect Storage via Hammerspace.   

Enjoy this on-demand webinar, “AI Data Pipeline Architectures” with Matt Fornito (“The AI Guy”), and Hammerspace Field CTOs Brian Bashaw and Chad Smith to learn about:

  • AI data pipeline architecture best practices
  • How to move data to compute effectively
  • How to overcome data silos
  • How to leverage existing infrastructure investment in new AI initiatives

Catch up on the latest episodes from the Data Unchained podcast

Subscribe to the Data Unchained podcast here, and listen on your favorite podcast platform, including:

Hammerspace Events in March

Visiting NVIDIA GTC 2024 March 18-21? Stop by Hammerspace’s booth #1718 or set up a meeting here

Discover the latest information and resources on Hyperscale NAS from Hammerspace:

Looking for additional ways to stay up to date about all things Hammerspace? Find us on the channels and media you love most, and don’t forget to like and subscribe!

Hammerspace’s YouTube Channel
Hammerspace on LinkedIn
Hammerspace on Twitter
Hammerspace on Medium
Global Data Environment Blog
Data Unchained Podcast