Getty Images

Cohesity, Microsoft Azure bring OpenAI to backup software

Analysts believe the potential exists for generative AI in data backups, but new features shouldn't distract from core competencies and customer demands.

Cohesity has entered a deeper partnership with Microsoft Azure to bring the capabilities of OpenAI into the vendor's backup software.

The new partnership also brings Cohesity's backup-as-a-service and data protection offerings into the Azure cloud platform, giving multi-cloud customers a choice between hyperscaler titans Azure and AWS.

Cohesity plans to integrate its existing machine learning and pattern recognition capabilities into chatbots and other technologies from OpenAI, a generative AI vendor now in the cultural zeitgeist due to its ChatGPT and Dall-E tools.

Cohesity said specific use cases for OpenAI integrations include detecting abnormal user behavior within a backup environment and turning speech or text into statistical graphs on demand, among others.

"This isn't AI washing -- this is AI-based [data] searching," said Sanjay Poonen, CEO and president at Cohesity, during a press briefing.

Microsoft is helping to keep the lights on at OpenAI with a multibillion-dollar investment over the coming years to grow the company and its technology.

But backup market experts have mixed opinions on how introducing generative AI will improve the experience for IT staff.

The desire to chase a new technology before it's fully vetted for the enterprise is a common trait among backup vendors, said Naveen Chhabra, analyst at Forrester Research. Backups themselves fill an important, if unsexy, niche within the data center that can be diluted if companies overvalue new tools unrelated to backups and restores, he said.

No one knows the problem, but trusts AI for its solution.
Naveen ChhabraAnalyst, Forrester Research

The early nature of generative AI as developed by OpenAI, which has yet to find enterprise adoption beyond experimental uses, only compounds this market chase, Chhabra said.

"No one knows the problem, but trusts AI for its solution," he said. "Backup vendors altogether are running too far ahead to adopt the latest and shiniest feature on the market, but they're leaving the basic capabilities of the market unaddressed."

Azure dreams

Cohesity DataProtect, its backup-as-a-service offering, now supports Microsoft 365 data and the ability to back up to Azure.

DataProtect and other Cohesity products, such as Cohesity Data Cloud and Cohesity Cloud Services, now also support integrations with Microsoft Sentinel, a security information and event management platform; Azure Active Directory, an identification and multifactor authentication service; and Microsoft Purview, a data intelligence platform accessible to Cohesity customers through a BigID partnership.

Cohesity FortKnox, a SaaS cybervaulting tool, will also come to Azure in the coming months and is currently available in preview.

FortKnox previously backed up data strictly to AWS like Cohesity's other products. New compatibility with Azure brings Cohesity closer to addressing the real-world demands of customers that operate in multiple clouds, said Phil Goodwin, research vice president at IDC.

Goodwin said Cohesity's partnership with Azure is a deeper integration than its reseller agreement with IBM, which was unveiled last month. More backup vendors will make the effort to natively support multi-cloud environments in the years to come, he added.

"They haven't picked one camp or the other -- they're trying to serve both," Goodwin said. "We're seeing an expansion of [vendors] addressing multi-cloud."

Welcome to the machine

Cohesity executives said they've only experimented with OpenAI technology in their products for about two months and didn't expect to implement a feature or release a new product with OpenAI anytime soon.

OpenAI's potential to recognize patterns of malicious use or flag disruptive actions to data shouldn't be discounted, Goodwin said. Colder archival backups, a target for ransomware attacks, could be scanned by the technology to find corruption before a ransomware detonation, he noted.

"Once you have a deep connection to that data, you can find out more about how it is being touched [and] moved," Goodwin said. "That doesn't make it an automatic positive hit, but you can call it to the attention of the IT organization."

Potential applications for the technology are useful in the abstract, but Chhabra said backup vendors such as Cohesity should address more pressing customer concerns. Backups are only valuable if they can be successfully recovered, and technological developments should focus on deciphering the odds for a successful recovery from a backup and how to beat those odds.

"[These services] tell you how much you've backed up your data, but haven't told you how much you can recover," Chhabra said. "As a CIO, CISO [or] vice president of operations, what do I want to know?"

Generative AI -- and AI services in general -- also generate computational overhead, likely increasing the resource strain on vendors, which could then pass those costs on to customers, said Jerome Wendt, CEO and analyst at DCIG. Using new tools to identify ransomware or other issues in data is useful, but shouldn't come at the expense of core competencies.

"Companies need to stay true to what they're good at," Wendt said. "I still believe anything you can do to stop ransomware from preventing a recovery is a good thing."

Tim McCarthy is a journalist from the Merrimack Valley of Massachusetts. He covers cloud and data storage news.

Dig Deeper on Data backup and recovery software

Disaster Recovery