Cradlepoint adds machine learning to NetCloud service to facilitate cost control

The new capability learns from each customer’s user patterns to project their cellular use across wide deployments, enabling them to avert costly overages.

Boise-based WAN networking specialist Cradlepoint has announced new machine learning capabilities for its cloud-based NetCloud Service. While the ML enhancements are not exclusively focused on the retail vertical, Cradlepoint is emphasizing their application there because the announcement coincides with next weeks National Retail Federation [NRF] show.

Cradlepoint first introduced NetCloud in 2016, as networking began its transition to cloud management.

“When we introduced NetCloud as a cloud management tool, it was the start for us of a move away from a mainly hardware model to where cloud software licensed on a subscription basis became the focus,” said Donna Johnson, VP of Product and Solution Marketing at Cradlepoint. “You still have to have physical hardware connection. But a lot of the value now is in the software. It has come to the point that we are really selling the NetCloud software, and the hardware comes with that.”

This announcement effectively marks the introduction of machine learning into NetCloud.

“This is the first time we have introduced proper machine learning, that goes beyond a straight-line extrapolation of data,” Johnson said. “It looks at past usage history unique to each customer, and draws conclusions based on that customer profile.”

What the machine learning does will be familiar to most from consumer products – except that it does it at enormous scale for the management of large distributed wireless WANs.

“All of our customers now are using cellular, and more of them use it as a primary link rather than as a failover, but with that comes a fear of exceeding their cellular data plans and getting expensive overages,” Johnson said. “What we are doing with the machine learning is learning each customer’s data usage patterns on their wireless links to project cellular usage, and determine if and when they will go into overage in the current billing period. From the customer’s perspective, this removes a major risk of using cellular. It works both with a one-to-one plan, and with a pooled plan for multiple locations.”

The announcement is being made around the upcoming NRF event, although the technology’s use extends well beyond retail. It does have particular resonance in retail, however.

“It’s not specific to retail by any means,” Johnson said. “It’s useful to any organization that uses cellular. But we are seeing the fastest adoption of cellular across retail, which has been very aggressive in adopting cellular.” Johnson noted that retail innovations as they adapt to a multi-channel universe include many more cellular WAN endpoints, in devices for next-generation display, marketing, shopping and loss prevention technologies, pop-up stores, kiosks and mobile showrooms.

“Retail was an early adopter of SD-WAN, but you can’t look at the way retail is going and expect that the same network model will work,” Johnson said. “Machine learning is one example of that. As we get to 5G, the number of connected things will continue to grow. With thousands of things connected to the networks, and many data plans, you need to be able to manage at scale. SD-WAN was built for a static broadband  world, but with cellular and 5G, and more pop-up locations and on-demand applications, the network needs to be a lot more portable, with the ability to segment out. Cellular provides a way to do that with connections that don’t share a WAN link, and that requires a rethinking to the SD-WAN paradigm.”

Cradlepoint has both reseller and managed services partners, and while the new ML capabilities are of advantage to both, they are particularly important to MSPs.

“They need the ability to automate tasks at large-scale, so this is important for them,” Johnson said. “Machine learning also enables all partners to offer new services around it as well.”

Leave a Reply

Your email address will not be published. Required fields are marked *