7 Storage Considerations for Enterprise AI

Published on June 17, 2021, this report was written by by Ken Clipperton, lead analyst, storage, DCIG LLC.

Until recently, data-intensive analytic workloads were mostly the domain of large-scale research and educational institutions. However, many commercial enterprises are now adopting these AI-driven workloads to enhance the value of their business data. As organizations embrace these workloads they should be aware of 7 storage considerations for enterprise AI.

AI in Enterprise
New AI and ML workloads place new demands on enterprise infrastructure. File-based workloads are at the heart of much AI-led innovation. As a result, unstructured data-including video, images and audio files-are expanding rapidly.

However, enterprises also seek to create new business value from consumer preference analysis and create products leveraging new levels of automation. As a result, datasets comprising structured data-including call records and credit card transactions-are also expanding. The optimal storage infrastructure should handle these emerging file-based AI workloads along with existing block- and file-based enterprise workloads in a single unified storage solution.

7 Storage Considerations for Enterprise AI
Enterprises expect many data management capabilities from storage solutions. The emergence of enterprise AI puts special emphasis on the following 7 considerations:
• Architected for consistent performance at scale
• Optimized use of multiple storage media to meet performance, capacity and cost objectives
• Unified storage through multiprotocol support
• Integrated enterprise data protection features
• Powerful management tools for simplified and automated infrastructure management
• Intelligent infrastructure analytics
• Experienced solution provider

Unified Storage Benefits
Storage that supports the concurrent use of multiple protocols enables the solution to consolidate a range of workloads, which achieves several business benefits:
• Makes data available to processes across the AI/ML lifecycle
• Eliminates the need to acquire, manage, and maintain separate SAN and NAS storage environments
• Enables consistent data protection
• Reduces acquisition costs, system management, and administrative overhead
• Minimizes data center costs for power, cooling, and rack space

DDN Brings its HPC and At-Scale AI Infrastructure Expertise to the Enterprise Under the Tintri Brand
Tintri’s parent company, DataDirect Networks (DDN), is a leading storage provider for at-scale AI and HPC environments and is, in fact, the largest private storage company in the world. DDN has developed an understanding of data-intensive workloads and the demands they place on storage at any scale. While many enterprises are just now exceeding the petabyte-scale storage threshold, DDN has clients with exabyte-scale deployments.

Over the past several years, as enterprises grew more interested in the potential value of data-intensive AI, DDN grew more interested in applying its expertise in data at scale to new solutions for enterprise customers. DDN’s acquisition of Tintri, Nexenta, and Intelliflash product lines, now collectively under the Tintri brand, has opened up access to thousands of enterprises and their broader enterprise infrastructure ecosystems.

These acquisitions also brought into DDN some of the most advanced AI and ML technology in enterprise storage, bolstering its Intelligent Infrastructure capabilities. Its broader strategy is to expand these and other AI-related capabilities and autonomous operations across the whole range of DDN and Tintri products.