EtherNet/IP Limitations

I’ve probably written more about EtherNet/IP than anyone in the industrial automation industry. I’ve been speaking, writing, and blogging about EtherNet/IP for over 20 years. It’s one of my favorite topics and something I am pretty intimately familiar with. I’m not as expert as our development team is, but I am a serious follower of all things EtherNet/IP.

Today I’d like to do something that I haven’t ever done before. I am going to give you my list of things that I don’t like about EtherNet/IP. I am going to talk about its limitations and when and where not to use it.

Let’s start with the most serious limitation, the CIP Object Model:

EtherNet/IP does have an object model. It has objects with attributes that are grouped into instances and classes. This has some benefits. An EtherNet/IP scanner always has a way to interrogate an adapter and learn about it. It always knows that there will be a set of required objects (Identity is the most important one) that it can interrogate to find out about the unit. Unfortunately, it’s not the kind of extensive object model that you find in OPC UA. But I give it a pass. At the time it was developed, all we had was the register and coils of Modbus. EtherNet/IP was a vast improvement over what we used before.

I have other limitations – things I’d like to complain about – but it wouldn’t be fair. I’m here about 25 years after the original designers started their work. With all the advancements of networking technology behind me, it would be easy to criticize a lot of what EtherNet/IP does, but that just wouldn’t be fair. I’ll leave my only complaint to be that object model.

Instead, let’s talk about where EtherNet/IP is today and where it is going in the future.

Many people today are hearing about Industry 4.0, the digitalization of the factory floor, and all that other stuff. It’s clear to me that some things are going to change. How much and how fast are some of the bigger questions under debate.

What won’t change for what I think is the next 5 to 10 years is the programmable controller as the central controller for the real-time control system. No matter what IoT Edge Computer you put out there, nothing will be as reliable, as well supported, or as appropriate for the task at hand today as Allen-Bradley ControlLogix or Siemens 12/1500 controllers. I don’t believe that anything else can do the job of running a machine as well as today’s sophisticated PLC.

Now, a lot might change around them. We might have other kinds of networks. We might have a mass of edge controllers performing all sorts of ancillary tasks, gathering data, executing some logic; but the core function of running a manufacturing process will always be the PLC, and that’s not going to change in the foreseeable future.

And EtherNet/IP and PROFINET IO will be the networks that those controllers rely on to move I/O between the controller and endpoint devices. Now those endpoint devices might have secondary ports to support an IoT protocol like MQTT or OPC UA, which is entirely possible. I just don’t see those kinds of networks taking over the core control function of a manufacturing network.

I started off this article intending to talk about the limitations of EtherNet/IP. I did that – at least one anyway. But in the end, despite their limitations, EtherNet/IP (and PROFINET IO) are here to stay as the center of our networking strategy on the manufacturing floor well into the 21^st century.

PS – For more information on EtherNet/IP, you should try my book: EtherNet/IP: The Everyman’s Guide to EtherNet/IP