Edge computing devices are proliferating at an astonishing rate, jumping from about 15 billion devices today to about 55 billion by 2022, according to Rob High, VP and CTO of IBM Watson.
“This marketplace around edge computing is enormous,” he said. “We don’t really know where the saturation point is … There will be a limit at some point, but we don’t know what that is.”
As edge devices continue to hit the market, “the No. 1 thing you begin to realize is that this industry is at risk of imploding on itself if it does not solve the problem of creating a standard way of managing it, [and] creating a set of standards that developer communities can begin to form and create ecosystems from,” High said.
“We think that the only way to achieve that standardization in a meaningful way is through open source,” he said. “Everything that we’re doing is based on open source, and it’s based on that merely because we don’t believe that we or anybody can be successful, both in the industry and at large … without building strong, vibrant, healthy ecosystems around standards.”
As such, High argues that it is paramount for the industry to foster the advancement of edge computing by reducing administrative costs. “That’s the thing that we have to tackle,” he said.
The good news is that edge computing doesn’t always require the installation of new equipment. Many organizations already have the necessary equipment in place — oftentimes supplied by vendors — and factories are particularly well positioned because they’ve already been installing IT equipment designed specifically to run workloads, High explained.
“The IT equipment that is already deployed on the factory floor, in hotel buildings, in retail stores, etc., is good enough,” he said.
IBM Targets Mutual Benefits of 5G, Edge
However, while many enterprises want to enable edge computing in remote or on-premises locations, wiring all of that peripheral equipment required to deploy edge computing “could be quite daunting if you have to lay wires all over the place,” High said. That’s where mobility, and specifically 5G, comes into play because it can enable use cases that require connectivity with equipment or remote locations, he added.
“The edge is accelerated by 5G, and 5G brings advantages to the edge computing scenarios. To some extent, I think 5G depends on edge as well,” High said, adding that 5G and edge computing are also addressing two distinct and interrelated problems.
5G and edge computing are advancing on parallel tracks with customer demand rising in kind, but many of the early enterprise dependencies on edge computing are occurring without the assistance of 5G today, according to High.
One of the biggest benefits of 5G — lower latency — will ride on edge computing. While 5G promises lower latency, “there’s a little bit of a fallacy in that promise” because the “end-to-end round trip times from the device up to the cloud is still about 500 milliseconds,” he said.
That latency isn’t due to the network or radio access network (RAN) equipment, but rather the laws of physics, the speed of lights, and inherent differences between the device and the cloud, High explained. Backhaul also remains a bottleneck in 5G, and the “only way to solve that is by moving workloads closer to the edge,” he said.
“At the end of the day, the promises of lower latency and higher bandwidth for the enterprise application doesn’t really get fully realized until they start to deliver commercial compute capacity in their network facilities,” High said.
Edge Enables Better Use of Bandwidth
On that front, he expects more network operators to deploy mobile edge computing capabilities this year, including a mix of public and private clouds, depending on the use case and special requirements. Enterprises will also have to make better use of bandwidth by limiting the transmission of unnecessary data.
At least 98% of data generated by IoT devices and transmitted to the cloud “never really ends up getting used in any meaningful way,” High said. Some of that data should never be transmitted to the cloud in the first place. “Rather than ship all the data up to the cloud, people want to start processing that data closely so that they can truncate the data and send only the signal back up in the cloud.”
As more factories are equipped with an array of cameras, for example, it quickly becomes apparent how much bandwidth is wasted if all those video feeds are constantly sent back up to the cloud. “All of that, I think, is driving a first wave of adoption that’s largely about taking the existing workloads and migrating them out to the edge,” High explained.
IBM has a “long history in the world of hybrid computing and mobile cloud management,” and it views edge computing as an extension or other example of those environments, High said. “There’s a confluence amongst multiple trends that are all lending itself to expanding edge computing.”