A few years ago I first started hearing the term “edge” used in conjunction with networking. Although edge computing had been around for some time, the term edge started creeping into every discussion and every presentation at industry conferences.
Today, nearly every company we talk to at SDxCentral wants to tell us about their edge computing product. And even service providers like AT&T and Verizon are talking about their network edge as part of their marketing message.
This type of frenzy around the edge reminds me a lot of the early days of cloud computing when every company wanted to talk about their “cloud” solution and many industry groups spent time defining the cloud.
But what exactly is the edge? I’m not the only one asking that question. About a year ago my colleague Linda Hardesty tried to unravel the puzzle in this article. But she noted that in terms of network infrastructure the definition of the edge includes everything from base stations to small cells and data centers to routers and even switches.
Recently a group of vendors and analysts also tackled this issue in a “State of the Edge 2018” report. The report focused on the architecture for edge deployments and provided some guidance on the future of edge computing.
The report defined the edge in two ways: the infrastructure edge and the device edge. And it said that compute will exist at both locations coordinated with a centralized cloud.
Even vendors are trying to define the edge. In a July blog from Kevin Shatzkamer, vice president of enterprise and service provider strategy and solutions at Dell EMC, he said he breaks edge into two different categories. The first group is the access edge, which is a terminating point on the network like SD-WAN or an IoT device. The second group is the network edge, which is the aggregation point within one network such as the data center initiatives like the Central Office Re-architected as Data Centers (CORD) and edge clouds.
Interestingly, Shatzkamer also groups edge into different use cases. For example, there is content at the edge; security at the edge; IoT at the edge; and data processing, or analytics, at the edge.
For David King, CEO of Foghorn Systems, which makes an analytics software that works on the edge of the network, the hype about the network edge has accelerated dramatically since the company first announced its funding in June 2016. “Everybody had overlooked the edge when we launched. Now everyone has an edge,” King said.
But King discounts those that claim even sensors on an IoT network are part of the edge. “Sensors don’t have context or awareness,” he said, adding that he believes that for something to be the network edge it has to be “close to the data but high enough in the topology to make it valuable.”
Jason Anderson, VP of business line management at Stratus Technologies, which makes an edge computing platform specifically for industrial applications, agrees that people are getting confused when they see IoT as the edge. “It’s a layer above the IoT device,” he said. “Most consumer IoT applications don’t need an edge computing layer. Where you need that additional computing layer is when the data is critical and needs protecting.”
Similar to the cloud, there are a lot of different approaches to the edge. And because of that it’s difficult to say exactly where the edge begins and where it ends. But like any new technology, the industry will eventually start to coalesce around a definition. And when that happens, we will likely start to have an easier time distinguishing the hype from reality.