Emerging Trends and Players in Object Storage for AI

As AI moves into the enterprise space, attention is focused on the quality and security of data. After all, AI inferencing, the process of running corporate data against pretrained models, requires good, well-organized data.

And there’s no shortage of that commodity. Companies everywhere have more data than ever before, gleaned from a variety of sources. Much of this data is unstructured, meaning that it doesn’t conform to the manageable schemas of traditional databases. Instead, it comes from documents siloed across distributed environments, images and video items captured in network attached storage, and even email, photos, and social media posts stored on desktops. By most accounts, this unstructured data comprises at least 80% of a mountain of data that’s growing exponentially.

Enterprises are finding that this data stored across a variety of locations and in a range of formats is increasingly valuable to enterprises because it holds unique information not accessible anywhere else. The issue is getting it all into a single location, ready and secured for analytics, business intelligence, and use in AI applications.

Thankfully, demand for unstructured data management for AI workloads is being met by a variety of companies, including longstanding providers of storage systems, software, and services who are adapting their tried-and-true technologies to fit the AI era in a variety of ways. This includes ensuring that data is easily accessible, organizable, and ready to feed to AI infrastructure based on GPUs.

In a new special report, Object Storage in the AI Era: Emerging Trends and Players, Futuriom takes a detailed look at the market for object storage products and services, parsing the trends that are emerging as a result of the need for unstructured data management.

Key Highlights

This reports includes information about customer requirements and how object storage technology is meeting those needs.

A few of the highlights covered include the following:

AI is changing the game plan for object storage providers. As inferencing takes hold in enterprise environments, vendors with longstanding expertise in object storage are expanding their platforms to address AI workloads.
The rise of object storage has invigorated global file systems. Vendors providing systems that furnish a common namespace for files and objects stored in various media and distributed geographic locations have also reached a turning point toward AI.
Unstructured data is emerging as key to AI applications. By most accounts, unstructured data—including text file, videos, audio clips, images, and other unformatted data—accounts for 80% of corporate data, comprising a rich source of input to AI models.
Unstructured data is held captive in enterprise silos. Unformatted data has often been housed in servers and on network attached storage (NAS), making it hard to retrieve for use in AI.
Object storage is ideal for unstructured data. By storing massive amounts of data in a flat file format requiring no hierarchical directory system, object storage is an ideal medium for storing unstructured data.
Object storage services are on the rise. Providers of content delivery networks (CDNs) as well as independent managed service providers are offering object storage services designed to compete against offerings from the leading public cloud hyperscalers.
AWS S3 is pervasive. AWS set the bar for cloud-native object storage nearly two decades ago, and its Simple Storage Service (S3) continues to govern the format of object storage for providers throughout the market, which universally offer S3-compatible REST APIs for use with their platforms.
Object storage must be adapted for use in AI. While object storage offers unlimited scalability, issues of networking, throughput, load balancing, and adaptability to AI workloads must be addressed in order for market needs to be adequately met.
Companies mentioned in this report: Akamai, Alibaba, AWS, Backblaze, Cloudflare, Cloudian, CTERA, DataCore, DDN, Deft, Dell, DigitalOcean, Fastly, Fujitsu, Gcore, Google Cloud, Hammerspace, Hetzner, Hewlett Packard Enterprise, Hitachi Vantara, IBM, Komprise, Leaseweb, Lenovo, MinIO, Microsoft, Nasuni, NetApp, Nutanix, Oracle, OSNexus, Panzura, Pure Storage, Quantum, Qumulo, Scality, Seagate, Storj, Supermicro, VAST Data, Veeam, Wasabi, WEKA, Zadara