How Industrial Edge Fuels Real-Time IoT Processes
Even a few years ago, data-intensive tasks still largely relied on costly cloud services or on-premises data centers.
According to estimates, 2025, three-quarters of enterprise-generated data will be created and processed at the edge – outside a traditional centralized data center or cloud. That’s up from just 10% in 2018.
But as a greater number of Internet of Things-enabled, data-intensive tasks take place on smaller devices, from tablets and smartphones to smart speakers, the need for speed and high performance has grown at “the edge.”
Edge computing is a distributed computing architecture that brings compute resources closer to the users, data and devices that need them, improving response times and saving network bandwidth costs.
Industrial settings, such as manufacturing plants, have low-latency needs that are best fulfilled at the so-called industrial edge, said principal analyst Alex West of Omdia.
“We’ve seen beyond the local data centers, there is that need for low levels of latency, the single-digit milliseconds of low latency. What are the capabilities that enable it to communicate with the cloud?
Check out the Omdia report on Edge: “Industrial Edge Compute and the Future of Automation – 2020 Analysis.”
Processes that happen in milliseconds and involve lots of data require resources close to the data and devices that need them. Consider a bottling plant, which manufactures thousands of bottles per minute.
“The industrial edge brings in a number of benefits that can’t be achieved by the cloud, and one of those is latency,” said Alex West, principal analyst at Omdia. “Coca-Cola is processing 2,000 cans a minute. If you are trying to do any real-time monitoring, the cloud is not going to do it. Edge is going to be key for processing that data,” he said.
The cost of processing power has decreased immensely, enabling the edge.
But so too, devices are becoming more application agnostic.
“We’re seeing greater convergence around IT and OT. That is opening up opportunities for open standards, which helps with that development of edge-based soluitons to make this equipment more application agnostic. You’re going to have processing at the edge where devices are a piece of compute and the software is going to define … the automation.”
Industrial Edge Still Needs Cloud Computing
At the same time, West said, cloud architectures are essential to the industrial edge. In order for processes to be automated and secure, companies need to develop data training models. And these models, which help processes continually learn, should reside in the cloud, West noted.
“We talk about training applications on the cloud but then deploying and inferring at the edge,” he said.
West said that IT pros need to think about the kinds of processes they need to perform, and where those processes take place.
“If you are monitoring an asset to predict failure, that isn’t where you need low latency. You don’t have to have that [process take place] at the edge,” West said. But quality control might require industrial edge compute, he said. “If you’re measuring quality, you don’t want to get too far down in the process. In that case, you want to be able to run models at the edge to identify them quickly.”