“This is not an easy subject to talk about. There are so many protocols, and terms have evolved—even the way the word ‘Ethernet’ has been used,” states John Ditter, input/output product specialist with Germantown, Wis.-based automation components vendor Wago Corp. (www.wago.com). “To most people, the word ‘Ethernet’ includes the TCP/IP stack. But these are different things,” adds Dick Caro, principal of network consultant CMC Associates (www.cmc.us), Acton, Mass.
The most fundamental source of confusion may arise from having the International Standardization Organization’s (www.iso.ch) Open System Interconnection (OSI) Reference Model and its seven-layer protocols stack, as well as the four-layer TCP/IP stack, commonly called the Internet Protocol or Internet Protocol suite.
At OSI’s bottom is the physical layer, on which the data-link layer containing Ethernet sits. “Ethernet is a communications protocol for only the lowest layers of the communications stack: physical layer and data-link layer,” explains Caro. IP and TCP individually comprise the OSI stack’s respective third/network and fourth/transport layers. The transport layer also contains the all-important UDP, or user-datagram protocol. Next, respectively, in ascending order are the session and
presentation layers. At the stack’s seventh, or application, layer reside TCP/IP protocol suites such as hypertext transmission protocol (HTTP); simple network management protocol (SNMP); simple mail transfer protocol (SMTP), for e-mail; file transfer protocol (FTP); and others.
It’s all Ethernet
“However, people classify all the layers as Ethernet,” comments Mark DeCramer, Wago’s product manager for advanced electronics. “We get phone calls from people who are interested in Ethernet, but don’t know which protocols to use. They think the two are synonymous,” Ditter says.
But they’re not, because Ethernet basically describes the hardware layer for the network, while TCP/IP runs on that
hardware, he notes.
In the TPC/IP stack, which contains the same functionalities as the OSI model but in fewer layers, at bottom is the network access or interface layer that includes Ethernet. Next in ascending order are Internet and transport layers containing, respectively, IP and TCP/UDP. The top layer is the applications layer that contains protocol suites mentioned for the OSI model.
It’s at the OSI model’s and TCP/IP stack’s applications layer that confusion also arises. Ditter says this is due to so many protocols existing, because “everyone is trying to build a better mousetrap.” Sometimes, end-users utilize industrial Ethernet, but don’t know they’re using the TCP/IP stack, DeCramer adds. “They just know that they’re using cable that has RJ-45 connectors (used commonly for 10Base-T and 100Base-TX Ethernet) on it and components that are very familiar to them, because of their familiarity with (Ethernet-based) office networks.”
Potentially causing confusion is that TCP/IP isn’t required or used for all Ethernet networks. For example, DeCramer says that ModBus TCP and EtherNet/IP use the TCP/IP stack. But some protocols having motion-control focus or functionality—for example, Profinet, SERCOS III (Serial Real-time Communication System) and Ethernet Powerlink—improve performance and make themselves more real-time by not using the TCP/IP stack, he adds. DeCramer also notes confusion might arise because IP means Internet Protocol with TCP/IP, but Industrial Protocol with EtherNet/IP, of the Open DeviceNet Vendors Association.
Clearly, the sources of confusion are many. But remember that Ethernet and TCP/IP are not equals. TCP/IP sits on top of Ethernet. And the TCP/IP protocol suites, Ditter’s Internet-focused “better mousetraps” that are closest to end-users, reside at the top of the TCP/IP stack.
C. Kenna Amos, email@example.com, is an Automation World Contributing Editor.