It can’t be denied: Edge computing is quickly escalating in importance as industrial enterprises realize they need a way to manage the large amount of data now entering the equation with the increased adoption of the internet of things, especially as devices are evolving to become less expensive and more powerful. Look inside today’s industrial enterprise and you’ll see edge devices going beyond basic automation control to an expanding array of advanced computing tasks, the kind you might previously have associated with the cloud.
There are a few reasons for this. First, many organizations don’t have the connectivity throughout their operational technology (OT) infrastructure to support sending large amounts of data to and from the cloud. Organizations are beginning to discover that having computing power happen at the edge may be more practical and cost effective.
Even in situations where robust connectivity is available, there’s another hurdle. With all of that production data streaming in, you need an efficient way to manage it. You may have to parse the data to determine what needs to stay on-premises and what can go to the cloud. All of that computational heavy lifting can create a latency problem that negatively impacts real-time process management. Industry analysts have estimated that only about 40% of data belongs in the cloud. In my discussions with industrial and commercial automation technologists, that number is more like 15%, including performance data that is aggregated for historical analysis to support process and business optimization decisions.
So what are you going to do with the other 85%? This question is driving increased recognition on the part of industrial technologists that edge computing doesn’t end with process control; the possibilities go far beyond the traditional supervisory control and data acquisition (SCADA) and manufacturing execution systems (MES). This recognition is fueling the availability of next-generation "edge intelligence" systems that take edge computing to the next level.
This includes real-time analytics platforms that process and analyze data right where it is generated, at the edge. For applications like advanced monitoring and diagnostics, real-time machine performance optimization, and predictive maintenance, it only makes sense to perform this automated decision-making as close as possible to the process itself.
Some organizations are deploying second-generation IoT and edge platforms that go beyond simple gateways to provide more sophisticated device management capabilities. In some cases, these systems are tailored to specific categories of applications and/or industries.
Even though the new edge infrastructure architecture is not yet fully defined, it’s fascinating to witness the pieces of the puzzle coming into view. These new analytics and IoT device management systems are a clear recognition of this new kind of workload at the edge and the tremendous value potential that can be realized by the “intelligent edge.”
Notably, these solutions are typically created specifically for use at the edge. For example, databases are designed to scale without resorting to cloud-based big data schemes. This might involve local storage-and-forwarding designs, sending data intermittently to a central database. The emphasis is on data management, rather than running applications.
For industrial applications, it is equally important that the hardware and software be designed for use in edge deployments where IT skills may not be present, or in field deployments where no human is present, such as a natural-gas compressor station or a wind-energy facility. That means prioritizing systems designed for ultra-reliability and serviceability, including remote maintenance, diagnostics and repair.
Given the exciting developments in real-time analytics and IoT management, it’s safe to say we will see even more advanced data-driven capabilities pushed to the edge. The cloud is here to stay, but the edge is where innovation is just getting warmed up. Watch this space!