Let’s take a bite of the industrial IoT layer cake

I spent the first part of this week at Emerson’s annual users conference, where I heard about bridging the gap between operations and information technology. This so-called OT-IT divide has been something I’ve covered for the last few years, but Emerson devoted its entire 5-day gathering to the topic. I even watched actors role-play a scene in which a plant operations manager goes to an IT person for permission to install new monitoring software, only to be told “no” because it would open up a port on the network.

It was a good reminder that, in many ways, the two sides speak distinctly different languages. Even I — who have had several patient people explain to me the intricacies of process control networks — sometimes get tripped up. Subsequently, in a conversation with Emerson’s Mike Boudreaux, director of connected services, I got a crash course in how to think about the layers that comprise an industrial IoT model.

You can find such layers in an oil and gas processing plant, but also in building maintenance or even a smart city deployment. For those trying to talk about the IIoT space, I think these layers help articulate something that is deeply confusing to the IT guys, who tend to only see an IT and OT divide. Guess what? There’s one more layer in there to worry about.

Let’s talk about all of them.

The safety layer: In an industrial plant, these sensors are wired and built into the physical infrastructure of the plant itself. In a building, the safety layer typically encompasses the fire detection systems and hardwired alarms designed to protect people from dangerous areas or circumstances. A smart city might include some lighting, water, and emergency power elements in this layer. Generally speaking, these systems have to measure events and communicate them in milliseconds. There has to be high redundancy, and once you install the safety layer, you don’t touch it again.

The control system layer: This is a closed loop system that should not be connected to the internet according to process engineers. In a factory, these sensors track data and adapt automatically to any new information — and they must register changes within seconds. In a smart city, ideally much of the basic infrastructure such the gear used to track water safety in rivers or water treatment plants, would operate this way. In buildings, this is where the HVAC system and connected lighting would be found. Boudreaux likened the control system to cruise control on a car: a bit of tweaking may be necessary here and there, but ultimately this is a tested system that should operate on its own.

IoT applications layer: This is where IT, APIs, and a bunch of other fun stuff comes together. There is no closed loop here; no changes are pushed down to the equipment. Instead, they are pushed to a human worker or a dashboard. This layer relies on a secondary layer of sensors attached to machines that monitor those machines’ health and activity in order to predict necessary maintenance, detect anomalies, and more. While some data from the control system layer might make its way up to this level after traversing a gateway, the control data is not going to the cloud. Analytics associated with this layer are performed in minutes or data is stored for later analysis and for later use in training machine algorithms.

I don’t often make a distinction in my reporting between the control system and the IoT applications layer, which means I sometimes find myself speaking with operations guys using similar language but not actually talking about the same thing. OT monitoring is different from IT monitoring. One can happen in the cloud; the other will likely never make it there. Data from the control layer is highly proprietary and largely used by machines. It’s the layer of automation that actually manufactures the end result.

Data from the IT layer is used by people, not machines, and typically presents less of a risk to intellectual property. In conversations with Emerson, I learned that this is where the company’s partnership with Microsoft Azure is focused. Data gets sent to Azure for storage and cloud processing, including plant data, machine health data, and even data around things like worker location. From there, it is packaged for predictive maintenance, dashboards that workers use to monitor the health of various equipment — even for displaying results in augmented reality.

So when talking about industrial IoT, either on the IT side or the OT side, figure out which layer you are actually trying to talk about. It may not solve the cultural divide, but it should improve communication.