While not old, EtherNet/IP can certainly be labeled “mature.” The specification was first introduced over 25 years ago, and the first EtherNet/IP conformed product left the ODVA test lab a little more than 20 years ago this year. Since then, tens of thousands of words have been written detailing the inner workings of the protocol, and thousands of EtherNet/IP control systems have been deployed across the globe.

The majority of these reliable and well-functioning control systems are the product of testing and experimentation: tweaking what’s worked in the past, testing new functionality and experimenting with new infrastructure.

Until now, no one has attempted to collect and document the best practices for deploying effective and reliable EtherNet/IP control systems. The EtherNet/IP specification from ODVA is a descriptive, not a prescriptive specification. For example, network designers will find that the specification does not specify an architecture for the underlying Ethernet network. There are numerous types of Ethernet networks that fit within the boundaries of the ODVA specification, but many of them would be inappropriate foundations for EtherNet/IP control traffic exchanges.

Figure 1 – Priority 0-3 form the lower priority and 4-7 form the higher priority.

A new book, The Everyman’s Guide to EtherNet/IP Network Design, fills this gap and details 12 specific guidelines an EtherNet/IP network designer should use to create practical, optimized and reliable EtherNet/IP control systems. It is a unique resource of proven and time-tested technologies, architectures and recommendations control engineers can use to design and deploy EtherNet/IP control networks on the factory floor.

To provide a snapshot of what this book encompasses here are two, highly abbreviated, segments of our recommendations.

Recommendation 3: “Control System Messages Get Priority. Period!”

The most important traffic an EtherNet/IP control system conveys is control signal traffic. Control traffic should preempt any non-control traffic. The impact of even momentary traffic congestion on an EtherNet/IP network could be significant and should be minimized. Non-control traffic, even network management traffic, should defer to control system traffic on an EtherNet/IP control system network.

The foundation for an EtherNet/IP network is an Ethernet network, which is also the foundation for most current IT networks. Because they share the same foundation, some facilities try to mix more traditional IT traffic with control signal traffic on an EtherNet/IP network. That creates the possibility of contention and/or congestion on that EtherNet/IP network.

Contention (access to the network) is easy to avoid by never using shared network infrastructure devices (Ethernet hubs) and always using full-duplex communication. Congestion occurs when multiple messages need to be simultaneously transmitted over the same link. There is a brief message transmission delay while one message is temporarily stored in the outgoing message queue of a switch waiting for the transmission of the other message to complete.

Congestion is managed using priority. The priority of an Ethernet message is indicated by the value of the priority field in the Ethernet frame. There are eight possible priority values, with 0 indicating the lowest priority and 7 indicating the highest priority. These priorities are mapped to specific message queues in a switch.

Two priority queues are sufficient for EtherNet/IP networks: high- and low-priority. EtherNet/IP implicit message traffic should get assigned to the high-priority queue. All other traffic should be assigned to the low-priority queue. No exceptions.

Recommendation 9: “Right Size Your EtherNet/IP Network.”

Right sizing means properly architecting the broadcast domains of both the Ethernet network as a whole and the EtherNet/IP TCP/IP subnetwork that is superimposed on that Ethernet network. The broadcast domains for each are not necessarily congruent.

It is only when devices certified by ODVA to conform to CIP and the EtherNet/IP specifications use a portion of an Ethernet network to exchange CIP communications that we have in an EtherNet/IP network.

An Ethernet network where most of the traffic is not EtherNet/IP traffic is not classified as an EtherNet/IP control network. An Ethernet network containing non-EtherNet/IP devices is a blended network and is also not an EtherNet/IP control network. Blended networks are strongly discouraged for machine control systems.

Right sizing an EtherNet/IP network means architecting an EtherNet/IP network to exist on its own subnet, with its own broadcast domain(s). It means deploying an EtherNet/IP broadcast domain such that all broadcast traffic is only delivered to the EtherNet/IP network devices that conform to CIP and EtherNet/IP specifications.

Figure 2 – Production zone with four EtherNet/IP control networks.

The network in Figure 2 is such a network. It consists of two production system switches with a number of controller switches, each having a PLC and a linear I/O network. A VLAN limits the broadcast domain to just the I/O devices on the linear network. Another VLAN connects all the PLCs, HMIs, and other operational devices across each system zone.

There are many advantages of this architecture, including line speed I/O communication, identical I/O device addressing, easily transportable logic programs and low resource devices that only see address resolution traffic from other I/O devices. The VLANs are architected very specifically to restrict broadcast traffic to the individual I/O networks, to restrict broadcast traffic to the controller network and to restrict corporate broadcast traffic to the corporate network. Class C networks are purposely selected to simplify device troubleshooting, training and easy sharing of PLC programs.

Right sizing an EtherNet/IP network means architecting it for the optimal size of the control system applications that it supports, limiting the broadcast domain of the EtherNet/IP network to the control network devices, excluding devices that are not part of the EtherNet/IP control system and organizing the devices to achieve optimal efficiency and reliability of the EtherNet/IP control system.

Other recommendations include:

• Use Fully Switched, Full-Duplex Ethernet: Architect EtherNet/IP control networks so that there is no possibility of contention between devices for access to the control network.
• Maximize Control Traffic Throughput: Architect an EtherNet/IP Device Network to Minimize Control Traffic Transmission Delays.
• Use Unicast Communications: Unicast traffic is the better choice for EtherNet/IP network traffic. It is far superior to both Broadcast and Multicast communications. Unicast traffic comprises most of all EtherNet/IP network traffic.
• Implement Redundancy Only When Needed: Only implement redundancy mechanisms where and when the expected value from implementing them exceeds the costs associated with implementing them.
• Continuous Monitoring of EtherNet/IP networks: Continuously monitor an EtherNet/IP network behavior to improve network performance and decrease response times to network disruptions.
• Use One Exclusive Plant Backbone Connection: Architect an EtherNet/IP control network so that there is a single connection between the control system and the corporate IT plant backbone.
• Define a Well-Architected Address Space for EtherNet/IP Devices: A well-architected EtherNet/IP address space is easily understood and explain-able to plant team members. It is an address space that is easily implemented without extraordinarily complicated router and switch configuration. And it is an address space that is easily maintained by a controls department.
• Properly Manage VLANs to Optimize Network Performance: Define logical networks to segregate traffic and ensure that Ethernet broadcast and multicast traffic stay within the VLAN boundaries and that any unicast Ethernet traffic can’t be switched but must be routed between VLANs.
• Make Your IT Department an Ally Not an Enemy: The problem between the IT and Controls groups is two very highly professional organizations that have good intentions but are separated by common objectives.
• Use Linear Network I/O Properly: EtherNet/IP network architects should use the 7 “rules of the road” when using embedded switch technology and linear networks in EtherNet/IP systems.

Get a Free Book

To learn more about these recommendations and building effective and reliable EtherNet/IP control systems, you can get the book, The Everyman’s Guide to EtherNet/IP Network Design, using the link below.

Fifty (50) free books are available to the readers of this publication. Simply click on www.rtautomation.com/IEB/freebook to get the book at no cost.

John Rinaldi is Chief Strategist and Director of WOW!, Real Time Automation (RTA)

Gary Workman is a coauthor of the book. Gary led much of the effort to identify and codify into practice the principles GM used to architect hundreds of reliable EtherNet/IP control systems at GM plants around the world. Each recommendation is followed by a “Gary View”, containing a personal, sometimes acerbic, story from his days fighting the GM IT department.