Storage

How automation and artificial intelligence could transform backup software

HAL-like control of data backups may be science fiction, but some of the basic concepts are not so far removed from reality. The person in charge of strategy for Commvault is intrigued by these ideas.

Image: iStock

It's a bad news, good news situation for system administrators worried about the reliability of backing up data from cloud computing systems.

The bad news is backup software for cloud data is only about half as mature as traditional versions, Commvault chief strategy officer Chris Van Wagoner said. The good news is some exotic technologies such as artificial intelligence may bridge the gap.

"I don't want to poke fingers at any cloud providers because I don't care who you are—they all experience outages from time to time," Van Wagoner observed, reflecting on the recent major outages at Amazon and Microsoft.

"One of the areas that everybody can do, including us, in the old legacy [backup] world is cross-platform, cross-application, cross operating system," he said. But for cloud data, "The ability to be portable between these different formats in my mind is a maturation process."

SEE: Video: Why the cloud is about more than just backup and recovery (TechRepublic)

Commvault is expanding its products for such portability later this year, although all backup providers must rely on the Amazons and Microsofts of the world providing the necessary programming interfaces and inside visibility, Van Wagoner noted.

Van Wagoner said he agrees that artificial intelligence could be a useful tool in battling that complexity. To be clear, "No backup software has artificial intelligence in it today," he said.

Where could AI start to have value in backup software?

"The interesting part is in AI's ability to be predictive," he continued. "I think within a decade, if you and I are still around, we won't be talking backup anymore. We'll be talking about how do we capture data, collect it, and use it."

To start, part of Commvault's focus for the next few years will be what Van Wagoner called "automation and orchestration" of cloud data backup. It's important because there are more moving parts compared to conventional network-attached storage and storage-area networks residing in your own data center, he explained.

Commvault will emphasize automation of repetitive tasks in service packs 8, 9, and 10 of its current software this summer, fall, and winter, Van Wagoner explained. Orchestration is more difficult, so there will be basic capabilities at first. Customers in government, healthcare, and Internet of Things sectors will find initial uses for this, he said.

Behind the scenes, Commvault is starting the longer-term effort by using machine learning, search, and workflows from a technology from Lucidworks. "We've talked about where we want to go from a high level," Van Wagoner said. "The idea of a job-based batch process is ultimately going to be replaced by something smarter and more efficient, much more real-time, and much more granular."

SEE: Research: Companies lack skills to implement and support AI and machine learning (Tech Pro Research)

Automation and orchestration may be a bridge to artificial intelligence. "I think within a decade if you and I are still around we won't be talking backup anymore. We'll be talking about how to capture data, collect it, and use it," Van Wagoner said.

IBM, through its Watson initiative, and Veritas are said to be evaluating similar ideas. Officials of those companies were not available to comment.

I wouldn't erase that folder, Dave

Artificial intelligence is already becoming part of storage in other ways. A startup called Ayasdi—which apparently is Cherokee for "search," though officials joked that it also means "available domain"—recently used its DARPA-funded analysis software to find unexpected striping failures for an unspecified hard disk manufacturer.

"In the early days of SSDs they would wear out so quickly... likewise with classic magnetic disks," explained Ayasdi's Ronaldo Ama, vice president of engineering. Drive manufacturers have preconceived notions of what can fail in a laboratory, so they weren't looking for other complications arising from real-world usage, he said. Ama previously worked at EMC, and he now realizes that such major storage vendors could use AI to test drives before they're ever deployed in modern petabyte-scale arrays.

Beyond backup companies, artificial intelligence in your data center could even protect systems by detecting an administrator's emotions. Affectiva is another startup using facial, skin, and speech detection to create responses in business applications such as human resources and safety coaching. The same functions could be applied to IT security, evangelist Boisy Pitre explained.

It's not so unrealistic that one day soon, if a Linux systems administrator has speech input enabled at her terminal, then AI would know if she's joking when she says Sudo rm -r out loud.

Also see

About Evan Koblentz

Evan Koblentz began covering enterprise IT news during the dot-com boom times of the late 1990s. He recently published a book, "Abacus to smartphone: The evolution of mobile and portable computers". He is director of Vintage Computer Federation, a 50...

Editor's Picks

Free Newsletters, In your Inbox