Getty Images
Storage vendors work with Nvidia to broaden AI portfolio
Building a relationship with Nvidia is critical for storage vendors that want to support data on premises, and they showed up at GTC with partnership stories to tell.
With all eyes in tech on AI, storage vendors are boarding the Nvidia train, with some vendors focusing on key areas and one attempting to touch everything AI.
At Nvidia GTC 2024, storage vendors including Weka and NetApp launched new offerings in an effort to hitch their cars to Nvidia, a maker of in-demand GPUs that underpin AI workloads. One noteworthy example came from Dell and its partnership with Nvidia to build an "AI factory," a technology stack that includes storage and GPU-based servers for AI model training and data services.
Storage and on-premises infrastructure vendors are showing customers they are prepared to support AI initiatives through partnerships and certifications with Nvidia, according to Brent Ellis, an analyst at Forrester Research.
"There is a perception that AI requires new storage capabilities, better performant storage, more storage," he said, adding that this is more marketing and corporate bragging rights than reality right now.
Partnering with Nvidia means storage vendors must meet throughput capabilities and address bandwidth and latency needs for AI, Ellis said.
Scott Sinclair, an analyst at TechTarget's Enterprise Strategy Group, said just about every vendor is working with Nvidia, including the major cloud providers. But these relationships have become especially critical for infrastructure companies that want to support keeping customer data in the enterprise.
"The partnership is important for everybody, but more important for on-premises providers," he said.
On-premises providers are seeing a bit of a renaissance when it comes to AI workloads, Sinclair said. In its report "Navigating the Cloud and AI Revolution: The State of Enterprise Storage and HCI," Enterprise Strategy Group found that companies are better understanding their storage options, including those for on premises. In that report, 78% of the 375 IT professionals in North America and Canada surveyed indicated that their organizations prefer to keep their proprietary, high-value data in their own data centers.
Legacy supporting cutting edge
At GTC, several storage vendors highlighted new offerings aimed at AI. Vast Data is helping to create new AI architecture that uses data processing units to offload the burden from and increase the performance of Nvidia GPUs. Weka unveiled its first appliance, obtaining the vendor's SuperPod certification for Nvidia DGX systems and claiming a performance at 765 GBps.
Others addressed broader enterprise AI use cases. Lenovo, for example, extended its Nvidia partnership designed to power generative AI applications for enterprise and cloud uses. This includes new Lenovo ThinkSystem servers with support for the Nvidia HGX AI supercomputing platform. The servers also support the Nvidia H100 and H200 Tensor Core GPUs and the new Nvidia GB200 Grace Blackwell.
In a new partnership, Hitachi will create its iQ portfolio that combines Nvidia technologies with Hitachi Vantara's AI storage platform. Hitachi iQ's first offering will combine the vendor's file storage software and hardware with Nvidia H100 Tensor Core GPUs and Nvidia AI software. The offering will seek Nvidia DGX BasePod reference architecture certification. And Pure Storage released an Nvidia OVX-validated reference architecture enabling cost- and performance-optimized storage designs for AI use cases.
Each of the offerings attempts to address a use case such as speed in AI training or setting up storage optimally to work with Nvidia processing, Sinclair said. Dell, on the other hand, is trying to cover as many bases as possible by partnering with Nvidia to build an AI factory, which combines infrastructure and resources to streamline AI initiatives.
"What people don't expect with Dell is the level, breadth and quality of services that they're delivering around AI initiatives," he said.
From storage to factory
In this case, the Dell AI Factory combines Dell's upcoming PowerEdge XE9680 server and PowerScale all-flash F710 storage with its data lakehouse and retrieval-augmented generation and model training. The Dell components and services, many of which are available now, are combined with Nvidia's GPU and AI software. Dell is offering its AI Factory as part of its Apex as-a-service portfolio and on premises.
It also includes access to Dell's Professional Services for GenAI, which is aimed at helping customers reach business results faster. Generally, services focus on deployment, troubleshooting and repair, Sinclair said. But Dell is expanding its services to include setting up customer AI initiatives.
"If you buy this service, by the end, you will have all of your data properly prepped and ready to go for a training environment for that use case," he said.
Nvidia, the only game in town?
While storage vendors unveiled partnership after partnership, the AI tech landscape is evolving. Nvidia might be a market leader in GPU sales and related technologies right now, but it isn't the only vendor.
"When there is only one primary vendor, what you have is a technology, not a market," said Marc Staimer, president of Dragon Slaying Consulting.
AI, including generative AI, needs high performance. While Nvidia is making the most noise, there are other options, including those from established vendors such as AMD and Intel that provide high-end GPUs, as well as from startups coming to market to meet the demand.
But Nvidia is providing more than chips for AI initiatives, Sinclair said.
"Nvidia has done an excellent job with its libraries and its development environment," he said.
Developers will likely want to work with tools they are familiar with, and Nvidia's tools are well established, Sinclair said.
Adam Armstrong is a TechTarget Editorial news writer covering file and block storage hardware and private clouds. He previously worked at StorageReview.com.
