Smarter and smarter things are happening in places that are not within the confines of IT data centers or clouds. There’s action out at the “edge.” The edge can be a manufacturing floor, smart city, smart building, energy grid, oil rig, windfarm, dairy farm, planes, trains or automobiles. Living on the edge in industrial settings are pumps, turbines, valves, robotic arms and other manufacturing equipment. When these “things” are smart and connected, then the entire affair becomes the internet of things or IoT.
At Hewlett Packard Enterprise we’ve realized processing power is needed at the edge because in so many instances IoT data can’t be easily or cheaply uploaded to the cloud. Think of security or data sovereignty policy and how it can prevent data transfer, when real-time response can’t tolerate the latency to the cloud, and when the volume of data is so large it’s unsuitable for the available network bandwidth. In these instances and more like them, the cloud is too distant to be of use for processing IoT data — especially when you consider that the tens of thousands of sensors operating out on the edge are capable of generating more data than all other big data combined.
So when should you use the cloud versus processing IoT data at the edge? Few questions can help you decide. For example, how soon would you like to know if your asset (e.g., pump or engine) is going to catch on fire? Most businesses would say they would like to know immediately, so that corrective actions can take place and disaster can be avoided. Or, how soon would you like to know the object in the road just ahead of a speeding autonomous car is a child versus a plastic garbage can? All of us would do whatever we could to avoid hitting a child, while a plastic garbage can is easily replaced.
Now consider scenarios where much more data is required to render an immediate and urgent action. A sealed manufacturing floor needs constant monitoring for excess moisture or leaking air. Finding and flagging these potentially crucial errors in real time requires capturing and instantly analyzing massive amounts of data, on the spot. Get it wrong or fail to act quickly enough and errors can pile up, creating defective products — or worse, compromising the safety of line workers.
For an even more extreme example, think about the efficiencies of a process plant, up to 90% of total cost of ownership has nothing to do with the original purchase price of the asset; but it has everything to do with the energy required to keep it running. Collectively in the United States alone, for the refining industry, that could be the equivalent of $20 billion per year that’s lost from unplanned downtime. By bringing compute to the edge, we can quickly and economically predict impending failures, threats or manufacturing problems and improve business efficiency.
This is why we say adding intelligence at the edge accelerates insight. HPE is working with partners in industrial plant and equipment to build industrial IoT by equipping operational technology, a category of hardware and software that monitors and controls how physical devices perform, with real-time data acquisition, pre-processing, monitoring and visualization. There’s a lot more to say about the intelligent edge and industrial IoT applications and the innovations. I’ll have more to say in this forum, but I’d also invite you to follow me on Twitter @TomBradicichPhD, where I lay out more principles of the IoT and what I’m seeing first hand with end users.
All IoT Agenda network contributors are responsible for the content and accuracy of their posts. Opinions are of the writers and do not necessarily convey the thoughts of IoT Agenda.