VDURA, Exxact Team Up To Address AI, HPC Storage Needs

Earlier this year, MES Computing spoke with VDURA CEO Ken Claffey on the demands AI is placing on storage needs.

VDURA and Exxact Thursday announced they have forged a new partnership to deliver storage to keep up with the demands of AI and big data.

VDURA offers software-defined offerings, and Exxact provides GPU-optimized hardware.

The partnership has sprung forth a new offering from the two companies—VDURA’s parallel file storage software on Exxact hardware—combining VDURA’s Data Platform with Exxact Storage Cluster.

AI and high-performance computing (HPC) applications aimed at processing massive datasets, running complex simulations and training generative models place big performance demands on storage systems, Google said in a blog post last year when it unveiled its own AI and HPC storage offering, Parallelstore.

[RELATED: A Managed AI Experience? There’s A New Service For The Midmarket For That]

Some of the benefits of VDURA’s new storage offering include flash-first architecture, parallel data handling, scalability and flexibility in on-premises or hybrid environments, VDURA said in a news release.

The new offering is targeted at AI/ML teams, engineering simulation groups, manufacturers and organizations managing large and complex data pipelines, according to a news release.

“Customers want systems that simply work: fast, reliable, and scalable,” said Jason Chen, vice president at Exxact, in a news release. “Our partnership with VDURA brings the storage performance our GPU platforms demand, giving users a smoother, more productive experience across AI and HPC workloads,” Chen added.

“AI and HPC users are pushing the limits of what’s possible, and storage can no longer be the bottleneck,” said Samantha Clarke, vice president of channel and partnerships at VDURA. “Our focus is on coupling enterprise‑class storage resilience with GPU and AI performance capabilities, ensuring organizations can meet the increasingly demanding requirements of today’s data‑intensive environments.”

Earlier this year, MES Computing spoke with VDURA CEO Ken Claffey on the demands AI is placing on storage needs.

“With the advent of AI, if you’re doing AI at any scale, you need a parallel file system. That’s not my view. That's Nvidia’s view. If you look at their reference architectures, they’re basically saying you need fast storage, and that is defined as a parallel file system,” Claffey said, during the interview.

[RELATED: VDURA’s CEO Breaks Down New V5000 All-Flash Appliance]

A parallel file system, he explained, “takes the data and the metadata and separates them, so you have the metadata [that] goes over here, and your data goes to a cluster of file servers. So just like you’re building a compute cluster where the cluster is a function of ‘n’ number of nodes, with a parallel file system, the performance of the storage cluster can match it because you can just add more file servers as you go.”