Getty Images/iStockphoto

Generative AI tools for backup and DR still in early days

Backup software vendors have pitched generative AI tools for automation and as virtual assistants, but how these additions enhance enterprise workflows remains to be seen.

Data backup and disaster recovery vendors want their generative AI assistants to become your new enterprise best friend, but the benefits aren't clear even after almost a year of hype and sales.

These GenAI additions for backup and DR include some form of chatbot to automate tasks or reporting duties, with a smattering of other capabilities on the side, such as code generation or alert management.

Although potentially useful, many of these additions are still more about keeping a product in the conversation rather than meaningfully changing operations, said Jerome Wendt, CEO of Data Center Intelligence Group.

When people are talking about introducing AI [to their products], it's more hype than reality.
Jerome WendtCEO, Data Center Intelligence Group

"When people are talking about introducing AI [to their products], it's more hype than reality," Wendt said.

Backup and DR vendors will need to make a strong case for how their own AI twists do not compromise customer security and offer additional value, said Sid Nag, vice president analyst at Gartner.

"Everyone is trying to AI-wash," Nag said. "[But the AI's usefulness] depends upon the solution and what the [business] value is."

Generative buzz

GenAI is the latest evolution of machine learning (ML) technology, able to generate media and code by synthesizing massive amounts of data generated from logs, data lakes and user-generated content. The term and tech entered the wider public lexicon earlier this year with the meteoric rise of OpenAI -- a generative AI company financially backed by Microsoft -- and its ChatGPT chatbot.

Since then, tech companies of all stripes and market targets, including backup and disaster recovery vendors, have scrambled to integrate some variant of generative AI into their platforms and portfolios to secure funding. Early adoption also enables vendors to begin training their AI services on software policies, procedures and manuals.

The major cloud hyperscalers selling to the enterprise are hyping generative AI tools as well. These include Google with its all-inclusive Vertex AI platform, Microsoft Azure AI and its Copilot assistant implementations, and AWS Bedrock for large language model generation. Each service offers GenAI capabilities at varying parts of the enterprise technology stack and contributes to the AI gold rush as vendors build off those tools.

Backup vendors are joining the rush with products such as Commvault Arlie, with GenAI capabilities coming from its Metallic data protection SaaS, and Athena AI for the Varonis Data Security Platform.

Both these AI tools debuted in November, joining numerous other AI releases over the past year, including Druva's Dru, Cohesity's Turing and Rubrik's Ruby. None of these products, according to vendor spokespeople, interacts with or stores customer data outside the customer's own environment. They instead use metadata or customer-developed data sets, as well as the vendor's own policies, for data-handling protocols. This ensures no data leaves the customer's domain, which is important for heavily regulated industries or for following data privacy regulations such as the GDPR.

Common features shared among these tools are a chat-style interface for interacting with the AI, anomaly detection tools and written report generation.

Many of these products are built on existing GenAI tools, such as Bedrock for Dru and Microsoft Azure's OpenAI for Commvault. More important is how many of these tools from backup vendors are building on existing ML tools released years ago, according to Krista Macomber, senior analyst at Futurum Group.

"By and large, data protection vendors have already been using machine learning to detect anomalies in the backup environment that could indicate a ransomware attack," Macomber said.

But there is still opportunity for differentiation, she said. For example, Commvault's AI product can look for AI-developed malware and provides intelligent backup scheduling and load balancing.

Backup software could benefit from GenAI in testing environments for ransomware recovery or workload reliability, making testing possible at a scale previously unavailable to smaller backup or security teams, said Christophe Bertrand, practice director at TechTarget's Enterprise Strategy Group.

Many ML technologies are already in use by backups, so those capabilities have further opportunities to expand, such as through automatic API generation across software and clouds, Bertrand said.

"There's already a lot of ML and other solutions to help [connect with services]," he said. "If it's a tool that helps you get there faster and better, why not [use it]?"

Backups could also play an increased role in protecting the AI infrastructure itself, Bertrand said, with the duplicate data helping to generate further training and learning for AI. For the backup vendors, these capabilities could further differentiate themselves from being IT products to the C-suite as more comprehensive security or data platform tools. Those capabilities can further tighten the grip of a vendor around a client as more data is ingested, making a switch to a competitor difficult.

GenAI has to prove value for DR

Beyond asking the user what they're looking for or automating a customer's backup procedures, there's still a lot remaining unproven about how GenAI will improve data backup beyond automation, industry analysts said.

AI-powered malware will eventually dominate the ransomware market, which will require AI-powered tools to stop these attacks, according to Marc Staimer, president and founder of Dragon Slayer Consulting.

Generative AI is leading to an increase in phishing attacks by enabling threat actors to spoof employees in email messages, said Matt Radolec, vice president of incident response and cloud operations at Varonis. However, while GenAI in backups might help to detect and warn of security compromises in the leadup to or aftermath of ransomware attacks, frontline data protection still rests on users following good cybersecurity practices.

It's a similar struggle to the security industry, as enterprise users must remain vigilant with digital hygiene, ensuring passwords are updated and that no illicit files are downloaded on company hardware, Radolec said. Business managers and supervisors will also need to drill proper security protocols much more than IT itself.

GenAI can learn and help fight these attacks, he said, but the human element will remain important to choose appropriate reactions and understand the scope of the attack on the IT stack.

"Getting people to understand it's part of their day-to-day responsibility is the thing," Radolec said.

Bertrand added that he expects security and backup teams to start using GenAI as a natural defense against these attacks, as no human alone will likely be able to stave off AI-backed assaults.

"You should be using AI to fight AI," he said.

Beyond snuffing out ransomware, Staimer anticipates AI becoming a much more seamless form of technology that employees use in the future, akin to the spell-checker today, which will transform how work takes place. Similar to Radolec, he doesn't see the individual employee going away, but rather having additional capabilities.

Having the human element remain in backups to understand what data is important to protect and avoid the costs of unnecessary duplication, for example.

"Generative AI is going to have a real impact on white-collar employment," Staimer said.

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