Data storage infrastructure company NetApp Inc. said today it’s looking to accelerate innovation in artificial intelligence with the evolution of its “intelligent data infrastructure” platform.

Announced today at the company’s annual user conference, NetApp INSIGHT, NetApp Intelligent Data Infrastructure describes the company’s vision of a data infrastructure that’s geared for hosting the data and AI models required to put generative AI into practice.

The announcement came as NetApp unveiled its latest all-flash storage system, the NetApp ASA A-Series, which is optimized for block storage and designed to modernize organization’s storage systems both on-premises and in the cloud. In addition, it also unveiled a number of enhancements to its cyber resiliency tools, with the aim of boosting data security for customers.

With its vision for generative AI, NetApp is positioning the NetApp ONTAP operating system for unified storage as the ideal platform for running AI applications of every type. To make this vision viable, it announced it has begun the certification process required to ensure NetAPP ONTAP can run on Nvidia Corp.’s DGX SuperPOD AI infrastructure.

This is a key step, NetApp believes, since Nvidia’s DGX SuperPOD AI has become the platform of choice for hundreds of companies building generative AI models, due to the high performance of that company’s graphics processing units. By certifying NetApp ONTAP to run on Nvidia’s DGX SuperPOD AI platform, customers will be able to take advantage of NetApp’s advanced data management capabilities, addressing the problems of managing the vast datasets required to train large language models.

The certification process is just one step NetApp has made toward advancing its vision. In addition, the company also announced the creation of a global metadata namespace that will allow customers to explore and manage data in a compliant way across their multicloud estates, paving the way for secure feature extraction and data classification for AI.

Meanwhile, it has directly integrated an AI data pipeline with ONTAP, enabling customers to automate the process of making unstructured data ready for AI. The data pipeline will perform key tasks like policy-driven data classification and anonymization, generating compressible vector embeddings to represent that data, before storing it in a specialized vector database so it can be accessed by AI models for training and retrieval-augmented generation-based inference.

Another key element of NetApp’s vision for generative AI’s data infrastructure includes a disaggregated storage architecture that enables full sharing of the storage backend while maximizing network bandwidth and the speed of its flash systems to increase performance and reduce the costs of LLM training and inference.

“This architecture will be an integral part of NetApp ONTAP, so it will get the benefit of a disaggregated storage architecture but still maintain ONTAP’s proven resiliency, data management, security and governance features,” the company said.

In addition, NetApp said, it’s introducing new capabilities for cloud-native services that will help to accelerate AI innovation in the cloud. These are meant to help customers create a centralized data platform for ingesting, discovering and cataloging their data for AI projects, which can be integrated with third-party data warehouses, the company said. It’s also developing a data processing service that will help customers to visualize, prepare and transform their data for AI projects. The infrastructure will make it simple to share these prepared datasets in a secure way, so they can be used across multiple cloud-based AI development platforms and machine learning systems.

NAND Research Inc. analyst Steve McDowell told SiliconANGLE that NetApp’s vision of an intelligent data infrastructure shows that it clearly understands the complex challenges faced by enterprises looking to deploy AI system. Moreover, he said it has focused substantial engineering resources on solving those challenges, and it’s promising some innovative solutions.

“The features NetApp is promising take clear aim at upstart competitors like VAST Data, which has been taking market share at the fringes of the market for high-end AI training data storage,” McDowell said. “NetApp is increasingly incorporating data manipulation capabilities into its data platforms.”

The analyst explained that NetApp’s vision is bad news for Vast and other competitors, because NetApp is well respected and has already established a presence in almost every large enterprise. “It’s the storage company with the most longevity, and it understands how to take care of enterprise customers,” he added. “So any enterprise looking at Vast will step back and wait to see what NetApp ultimately delivers.”

In addition to taking aim at the smaller storage startups trying to disrupt the AI market, NetApp is also striving to catch its traditional competitors flat-footed, McDowell believes. He said that none of its competitors, bar IBM Corp., have the resources or wherewithal to attempt to do what NetApp is trying to do. However, he warned there are no guarantees that the company will be able to deliver on its vision.

“Announcing all of this so far ahead of delivery is uncharacteristic for NetApp, and so it’s putting a lot of pressure on itself to deliver,” McDowell explained. “It’s a big promise to execute against, especially with the entire storage market watching.”

New flash hardware

As for the new all-flash NetApp ASA A-Series storage arrays, these are said to provide three crucial benefits to customers looking to manage block storage workloads, where data is divided into blocks of equal sizes and stored on physical media for rapid access.

The primary benefit is simplicity, with customers able to leverage a single storage solution that can be deployed, managed and upgraded with a few simple clicks. The company says deployment takes just a couple of minutes, and then the servers can be provisioned in mere seconds, with day-to-day management automated through its integrated AI operations technology.

Designed specifically for VMware and database workloads, the ASA A-Series delivers what NetApp says is “abundant performance and efficiency,” with millions of input/output operations per second, subsecond latency and intelligent data management capabilities, plus advanced features such as scale-out clustering and workload balancing. Further, it’s designed with performance in mind, NetApp said, with a 25% to 50% lower upfront price and a significantly better return on investment over time.

What’s more, to support the launch of the NetApp ASA A-Series, which will become available next month, the company has rolled out an update to its Data Infrastructure Insights service, which used to be known as Cloud Insights. The rebranded service is said to improve the server’s data monitoring and analysis capabilities, so customers can optimize the performance of their systems more easily.

McDowell was impressed with the ASA A-Series array. “The common wisdom has been that hybrid-arrays are dead, but NetApp is proving that wrong,” he said.

The analyst explained that there’s still a big appetite for flash performance in the nearline storage market, but the problem is that customers are very cost sensitive in this area. While almost every storage company offers a QLC capacity-optimized product in this segment, those arrays can be expensive. So the hybrid ASA A-Series array gives customers an option that combines the best of both worlds, delivering flash performance at a lower cost, he said.

“The magic that makes this compelling is NetApp’s embedded tier of software, which places data on flash or hard drive based on the performance needs of the workload,” McDowell said. “It’s a compelling option.”

Advanced ransomware protection

Finally, NetApp is releasing its previously announced NetApp ONTAP Autonomous Ransomware with AI offering in general availability starting today. This is a new, AI-powered offering that’s claimed to provide 99% accuracy in terms of detecting ransomware threats before they spring into action and cripple customer’s information technology systems, the company said.

Known as ARP/AI, the solution works by monitoring data workloads for abnormal activity, taking automatic snapshots of data from just prior to when the attack was launched, so that customers can recover their systems rapidly, without spending thousands or even millions of dollars on a ransom. The company promised that this machine learning-based service will be continually updated to ensure it can identify the most advanced forms of ransomware.

And in a related announcement, the company also revealed updates to the BlueXP ransomware protection service, which is designed to safeguard third-party systems. According to NetApp, this now integrates with Splunk Inc.’s security information and event management platform to ensure the safety of customer’s data.

International Data Corp. analyst Phil Goodwin said these updates will ensure NetApp is not only able to maintain data integrity, but also help customers to respond to ransomware attacks and restore access within just a few minutes. “These advancements in leveraging AI to enhance security directly within enterprise storage through its ARP/AI and data classification capabilities are key differentiators for NetApp,” the analyst concluded.

Image: SiliconANGLE/Microsoft Designer