As industry stalwarts like Matei Zaharia indicate that “the next promise in AI is in domain-specific enterprise use cases,” NetApp, a data management and storage provider, finds itself well-positioned to capitalise on this trend.
At the recent NVIDIA GTC event, CEO Jensen Huang highlighted NetApp’s role, stating “nearly half of the world’s files are stored on-premises on NetApp’s platform”.
With 76% of global tech companies and 70% of Indian enterprises having AI initiatives underway, outpacing the 49% global average, NetApp’s storage footprint could prove invaluable.
Notably, 91% of Indian companies plan to leverage over half their data to train AI models by 2024. NetApp is banking on this pace in the Indian market. However, its leadership isn’t surprised.
“India has always been strong on the tech side,” said Puneet Gupta, NetApp MD for India and SAARC, in an exclusive interaction with AIM.
He pointed out that India’s vast datasets, such as those generated by initiatives like UPI and Aadhaar, are crucial assets. He also said that India has the ingredients for AI success.
India, NetApp’s fourth-largest APAC market, has the potential to become the largest in the next few years, according to Gupta.
(Source: Polygon.io)
Globally, with a $26.48 billion market cap and Q4 FY2024 revenue of $1.668 billion, up 5.1% year-over-year, NetApp is poised for growth driven by all-flash arrays, cloud services, partnerships, and enterprise AI adoption, if executed well.
Partnering With NVIDIA to Be Enterprise’s Go-to Choice
NetApp has partnered with NVIDIA to advance RAG for generative AI applications in enterprise. It allows integration of NVIDIA’s NeMo Retriever with NetApp’s ONTAP storage, enabling LLMs to securely access the vast amounts of enterprise’s data stored on NetApp without compromising privacy and security.
Enterprises can leverage their existing data assets on NetApp to “talk to their data” through simple prompts and gain insights for generative AI, without the need to move data.
“The work that NVIDIA is doing in the AI space is remarkable,” Gupta said. “While they focus on building compute and server farms, we provide high-performance storage to make the whole solution work effectively,” he added
Moreover, the partnership complements NetApp’s existing AI services, which have been used by over 500 joint customers for AI model training and inference.
NetApp has also worked with NVIDIA to update its FlexPod AI converged infrastructure to support NVIDIA AI Enterprise software and was one of the first partners to complete storage validation for NVIDIA OVX systems.
This collaboration is significant because it allows enterprises to safely use their proprietary data within LLMs without the risk of data leaks or privacy concerns. It reduces friction, cost, and time to value for RAG by enabling access to data wherever it is stored.
Gupta also elaborated on the partnership with Cisco, particularly in the Indian context.
“We have a strong partnership with Cisco globally and in India. They work with NVIDIA as well, and we fit into the ecosystem by providing the necessary high-performing storage to complete the AI story,” Gupta explained.
Unified Data Management
Another critical challenge in data management has been the proliferation of disparate storage systems for different data types, such as blocks, files, and objects. NetApp addresses this issue by unifying all data services on a single platform, simplifying storage infrastructure and enhancing efficiency.
Shuja Mirza, NetApp’s India/SAARC director of solutions engineering, explained, “NetApp is doing its bit by ensuring that you can run all these data services from a single platform. So unifying all of it—objects, files, blocks, structured, unstructured—you know, the ability to store data and data lakes, put it to good use through modern workloads like Spark, Hadoop, etc.”
This consolidation supports modern workloads like Spark and Hadoop, which are vital for handling large-scale unstructured data.
Mirza emphasised NetApp’s commitment to providing customers with choice and flexibility, stating, “Our job as a technology provider is to make sure that we provide you with that choice. And, whichever format you’re getting data in, we’ll help you store and manage it efficiently, and make sure it is available to you all the time.”
Features like NetApp FlexClones allow data scientists to instantly create writable copies of datasets for experimentation without consuming additional storage.
NetApp also enables data governance frameworks to manage privacy, compliance, and access controls, which is vital when dealing with sensitive data used for training LLMs.
More Efficient Data Centers
With India’s data centre industry booming, to accommodate AI needs, including investments in GPUs and larger racks, Gupta sees significant opportunities for NetApp.
“The infrastructure needed to support the generative AI wave includes both large-scale cloud provider setups and enterprise-specific builds. NetApp aims to participate in both segments, leveraging our expertise to manage data efficiently across various platforms,” he stated.
NetApp could also provide storage efficiencies and a secure platform, including ransomware recovery guarantees.
“AI, especially generative AI, is all about having large datasets and creating copies of these data sets,” Mirza said. Adding that, “NetApp’s solutions help reduce these copies through snapshots, thus providing storage efficiencies and a secure platform.”
Mirza elaborated, “Typically, in these projects, the data scientists create environments. Each environment has got its own copy. So the volumes are large. With NetApp FlexClones it becomes easy and efficient by storing those copies through snapshots.”
This approach not only ensures efficiency but also secures data management, which is critical in AI projects. “Storing data is the tip of the iceberg. But if we were not to really do a good job of managing and securing it, then there is always a chance of breach and influence,” he asserted.
And even in case of a breach NetApp provides a 99.999999 guarantee with a recovery time of a few seconds.
The post How Does NetApp Use Half of the World’s Data in AI? appeared first on AIM.
