Every technology has a life span. It’s introduced. Innovators and early adopters get a hold of it. If it works, it gets accepted by the marketplace. There’s an early majority that gets in next, followed later by the late majority. Finally, the people that didn’t like it or who won’t take risks finally feel that it’s safe and they get in. They’re the laggards. Every technology follows this path.
Where is EtherNet/IP on the technology life cycle? It’s twenty years old this year. It’s getting a little gray and developing a little bit of a paunch, but it goes to work every day doing what it was designed to do. The early and late majorities have certainly adopted it. Even a lot of laggards by now have realized that to access a manufacturing system based on an Allen-Bradley, ControlLogix architecture you need to have EtherNet/IP.
Has it outlived its usefulness? Probably not. DeviceNet and Profibus DB are certainly on the wane, but the problem there is with the underlying technologies. DeviceNet is based on a CAN standard that is limited. CAN supports 8-bit frames, low speed, and bandwidth. In today’s architectures, we need to move a lot more data than I/O bits. When you do that over CAN, you get devices that monopolize already limited bandwidth networks. Profibus DP isn’t as speed limited as CAN is, but it is also limited by the frame size. It is also burdened by special connectors, ASICs (Application Specific Integrated Circuits) and PHYS (physical interface chips) that are needed to access the network. The best days for both these networks are behind them.
EtherNet/IP doesn’t have the limitations of DeviceNet and Profibus DP. It uses standard Ethernet which (currently) is the physical network of choice. There is no successor on the horizon for physical communications though I don’t doubt that one will appear at some time in the future. EtherNet/IP’s wide acceptance arises from the fact that it is a very good (excellent actually) network for moving I/O data between low-level manufacturing devices and controllers. And, we can’t forget, it is the standard used by the most popular controller in North America (ControlLogix).
If it has one weakness, it is how it interconnects with enterprise and cloud applications. It’s just not very good at that. It was designed in the 1990s when we were still struggling with things like Windows NT. The internet was around at that time but hardly useful. The very first browsers were being used at that time, but few if any of us could imagine how the internet would change the world and our manufacturing systems. EtherNet/IP was imagined as the successor to DeviceNet and ControlNet. It was designed to do what it does, I/O device communication, and do it very well.
The only tiny blip on the future radar for EtherNet/IP is OPC UA. Some in the OPC UA world imagine a scenario where OPC UA becomes the I/O standard. Indeed, I have advanced this case myself from time to time. The advantage of OPC UA as the I/O standard is that it offers some functionalities that EtherNet/IP does not. I’ll name two here:
Publisher/Subscriber operation – You can make a case that on a very fast Ethernet network (10Gig, 100Gig), cyclic communications don’t make a lot of sense. On a network like that, even extremely fast mechanical valves aren’t going to change in microseconds, so sending cyclic data that just repeats itself and uses bandwidth and consumes processing resources for no reason is just silly. An architecture where a device publishes data when the data changes make more sense. In that scenario, many more devices can more easily get access to the data than they can in an EtherNet/IP architecture.
Multiple Connections – It’s become advantageous for devices to also supply information like energy usage, diagnostic and repair data, and the like to enterprise and cloud applications. OPC UA can easily provide that information channel in a very standard way. With EtherNet/IP, that data has a much more complicated and frail mechanism to travel through the controller.
So, it’s possible that we may be on the cusp of the end of EtherNet/IP. I’m not predicting it and if it happens, it will likely be far off in the future.