In storage HPE also competes with Pure Storage and NetApp, both of which provide flash storage and predictive analytics technology. Flash storage, while more expensive per bit than hard disks, delivers data much faster while taking up less space and power.
HPE bought Nimble, whose stronghold was in flash arrays aimed at the midmarket, in part to complement 3Par flash technology aimed at higher-end enterprise systems. But the InfoSight predictive analytics platform was Nimble's "crown jewel," said Moor Insights & Strategy’s McDowell.
How InfoSight predictive analytics work
InfoSight collects infrastructure information from "call home" sensors and runs analytics against the massive amount of usage data it has accumulated over the years to detect patterns in order to predict, for example, when a user might run out of storage or when a storage device may exceed a certain input/output threshold. The sensors are in the storage devices themselves but also collect network, compute, and hypervisor data.
HPE says that the InfoSight update goes beyond alerting users to potential, specific system failures, offering recommendations to more efficiently provision network storage systems and improve performance.
"We’re watching what's going on in the environment and we’re putting it through these expert systems using quite a lot of different AI and machine learning and data science techniques; the goal here is to tell the customer what to do to make the environment better," said Gavin Cohen, vice president of product and solutions marketing at HPE.
"So, we might notice that everything's running just fine, things are humming along, servers are working fine, applications are working fine but there’s an opportunity to get more performance if you were to move for example a virtual machine from one server, where there's extra contention, to another place," Cohen said.
More and more data will be available for InfoSight's AI algorithms as InfoSight gets deployed across 3Par and other HPE systems. So as HPE brings InfoSight across its product line it will also lay the foundation for autonomous data center infrastructure, where system tweaks are made automatically by the system itself. "It's not a big leap," Cohen said. "The hard work is what we're doing now."
Moor Insights & Strategy’s McDowell agrees that the groundwork for AI-enabled autonomous systems has been laid, but says it's debatable whether enterprises will want completely autonomous systems. "I’m skeptical that IT wants anything autonomous," McDowell says. "The customer wants to be notified and make a choice; they’d be terrified if you're shrinking and expanding their data."
Sign up for Computerworld eNewsletters.