“When you think about a retailer that has 2,000 locations across the country, it’s too expensive to deploy on-premises data processing and analytics for every single location, so that’s where edge computing can be a huge boon,” says Paul Savill, senior vice president of product management and services at technology company Lumen, who points out that edge computing is designed to work in tandem with cloud. “Edge nodes combine hardware-driven computational power with software-defined networking capabilities to connect it to the public cloud,” he explains. “From one centralized node in one market area, say, the size of Denver, edge computing can serve many more retail locations within five milliseconds.”
Opportunities outweigh challenges
Shivkumar Krishnan, head of stores engineering at Gap, says the biggest challenge to making edge computing a reality in retail is legacy infrastructure. “As an end user on the cloud, it’s much easier to upgrade, since you can simply push a button and shut down or replace a virtual machine. In retail, it’s more of a logistics problem,” he explains. When setting up the first time, each location needs to connect its devices to the edge, which may need to be done at night, when customers aren’t in the store. And with vendors working on-site, store security staff as well as the manager will need to be on hand. “It really becomes more of a logistics challenge to figure out the availability of everyone,” Krishnan says. “And the process needs to be repeated for each of our 2,500 stores.” In the cloud, one push of a button can deploy hundreds of servers.
Data security is also an inevitable challenge when it comes to the internet of things and other digital devices. “The more you concentrate information in a location, the more you have to worry about protecting that, and the riskier that becomes in terms of creating a single spot that can be penetrated, and information stolen,” says Savill. But edge computing supported at nodes in nearby data centers and connecting to the public cloud are generally more secure and reliable than what a retailer could do on its own. That’s because edge providers, much like public cloud providers, are providing cybersecurity from a central location, on a mass scale, so they have visibility into what the threats are and how they’re affecting their customers, says Savill.
That said, the benefits and opportunities of the edge far outweigh potential challenges. “One of our biggest use cases for edge computing is at the point of sale, where we process millions of transactions,” Krishnan explains. From the store to the cloud, there are many failure points—switches, routers, the telecom circuit, and cloud providers. “The edge gives us a high level of redundancy to process all transactions at the store itself and fall back to the cloud if the edge fails,” he says.
“The edge gives us the full redundancy to process all transactions at the store itself and fall back to the cloud if the edge fails.”
Shivkumar Krishnan, Head of Stores Engineering, Gap
Gap has invested in edge servers over the past few years, says Krishnan, as part of an overall platform using the latest technologies such as microservices, cloud computing, streaming services, and a DevOps approach to engineering. “Now, with our platform, we can build, validate, and deploy applications with rapid turnarounds—all within the same day,” he says. “I can remotely monitor and manage the majority of our over 100,000 devices. Our sales associates use iPads that give us the ability to build native mobile user experiences that are intuitive.”
While Gap was early to the edge computing game, the challenge is keeping up with the latest and most advanced technologies, as with any technology adoption. Today’s edge servers have built-in graphic processing units, network routers, and broadband technology 5G, “all packaged in small-footprint devices that are built from the ground up for advanced machine learning,” he says. “Hopefully, we will catch the next iteration of these advancements and leapfrog others who get them now.”