Pure Storage is introducing new AI features, including a new copilot aimed at providing efficiency, ease of use and security, while expanding SLAs for data protection and recovery.
Pure Storage is focusing on two popular IT topics with its addition of automation and AI to its storage offerings, but it's also adding or extending service-level agreements that are aimed at performance and security.
At Pure//Accelerate 2024 in Las Vegas on Wednesday, the vendor expanded fleet-level automation of Pure Fusion, its storage-as-code offering, in an effort to unify an enterprise's arrays and optimize storage pools regardless of data type or location. The company is introducing an AI copilot for data management and data protection that's trained on its hardware and software usage. Pure has also expanded its SLAs to include more efficient GPU utilization as well as added a stronger recovery and resiliency agreement, a new security assessment, an AI-powered anomaly detector and site rebalancing.
Pure's news speaks to a larger trend playing out in 2024 as vendors work to establish a closer alignment between AI and storage as well as making its tech easier to use, according to Steve McDowell, founder and analyst at NAND Research.
"Storage is really complex," he said. "You need special skills to manage it, and what Pure is trying to do is simplify it."
Automation expands in Fusion
Pure's additions are focused on its portfolio, unifying storage and data management across hardware, software and subscription services. The platform uses Pure Fusion, and now the company is adding those autonomous capabilities to its storage arrays, according to Prakash Darji, vice president and general manager of FlashArray at Pure.
"We simplified management for a box a long time ago with Purity [Pure's operating system]," Darji said. "We now have customers with thousands of arrays and more data under management."
The latest version of Fusion is embedded within Purity and enables users to define fleet policies and fleet-wide provisioning at scale, he said. This new version is backward compatible with older arrays.
I think ChatGPT-like interfaces in front of your infrastructure software is the new normal.
Steve McDowellFounder and analyst, NAND Research
Along with automation, Pure added a generative AI assistant for storage. The copilot relies on hardware and software data insights from Pure customers to help investigate management and proactively protect data.
Generative AI assistants will probably become ubiquitous for complex, detail-oriented IT infrastructure including storage, McDowell said. He pointed to Dell Technologies and its introduction last month of a knowledge assistant to address infrastructure questions related to PowerStore.
"I think ChatGPT-like interfaces in front of your infrastructure software is the new normal," he said.
Feeding GPUs
Pure also said it expects to earn its Nvidia DGX SuperPod certification this year, joining other storage vendors such as NetApp, IBM and more recently Weka. SuperPod certification is Nvidia's data center reference architecture for high-performance computing (HPC) and AI workloads; it often serves as a signal that the storage partner can keep GPUs fully utilized. Once achieved, this would make the second Ethernet-based SuperPod architecture, the first being from Dell. Prior to Dell, SuperPod certification required higher-performance InfiniBand networking.
SuperPod certification isn't easy to acquire, McDowell said, adding that if a vendor wants to compete in high-end AI and HPC, certification is almost a requirement.
"Vendors really have to deliver, and what Pure is saying by getting this certification is that their stuff does," he said.
Separately, Pure is also releasing a new SLA for AI in Evergreen//One, the vendor's storage-as-a-service offering. The Evergreen//One for AI SLA guarantees storage performance to GPUs used in training, inferencing and HPC workloads.
Pure's FlashBlade//S500 can scale its throughput up or down to meet the new SLA and hit 100% GPU utilization, according to Pure's Darji.
SLAs for security, balance
Pure's SLA changes extend to three new agreements around cybersecurity, threat assessment and site rebalancing, where resources are rebalanced every year to better reflect a company's needs.
Customers have certain business outcomes and expectations that are prioritized over new technical features, according to Lara Greden, an analyst at IDC. Pure can set itself apart by focusing on SLAs and delivering on these agreements.
"SLAs address risk for customers," she said, adding that Pure is guaranteeing it will deliver against certain cybersecurity risks.
Pure's Cyber Recovery and Resilience SLA builds on its previous ransomware recovery SLA in Evergreen//One to now include disaster recovery. Customers receive a guaranteed recovery plan as well as a new cleanroom box, while keeping the previous for forensics. The SLA comes with a new security assessment that looks at fleet-wide deployments to check security policies and how they align with NIST 2.0 standards, a framework to help businesses assess and reduce cybersecurity risks.
Pure also introduced an AI-powered anomaly detection feature that looks for unusual or malicious activity. This new detection builds off the company's previous data reduction anomaly detection.
Greden said using machine learning for anomaly detection is becoming table stakes for IT vendors, be they networking, compute or storage.
"It's the same type of problem that you're aiming to solve: visibility into performance from an availability perspective and from a security perspective," she said.
Adam Armstrong is a TechTarget Editorial news writer covering file and block storage hardware and private clouds. He previously worked at StorageReview.com.
