I like the subtitle in the print version: “Industrial Ethernet Cruises Powerfully Along. But Can It Be More Than You Really Need? Are Simpler, Hard-Working Device-Level Networks a Better Choice?”
The “Ethernet Cruises” refers to the IMS Research study that I highlighted in a recent post: Fieldbuses, down; Industrial Ethernets, up? Yes, Industrial Ethernets are growing, but so are fieldbuses. (If you’ve already read the article, you’ve noted some differences in definitions: Dan calls process fieldbuses “fieldbuses” and factory fieldbuses “device-level networks.” I usually just say process fieldbus and discrete fieldbus.)
So, are simpler networks a better choice? No, but sometimes (ONLY sometimes) they are good complementary choices. I’ve always felt that you should start designing your system architecture with PROFINET, then add other networks if necessary. See Dueling Views of PROFINET’s Place in System Architecture. One benefit frequently cited by PROFINET users is the ability to reduce the number of networks needed! At the PROFINET Executive Leadership Forum last year an OEM told us they eliminated three or four other buses and just used PROFINET.
Dan quoted me saying, “Ethernet will work its way further down into simpler and simpler devices.” I’ve noted that it makes no sense to add an Ethernet connection to a simple sensor or actuator… today. Remember that there was a time when engineers had trouble imagining Ethernet connections down to IO blocks. I believe Ethernet will continue to make its way down the automation pyramid. (Do we still visualize a pyramid for automation? I guess I do.) Today there may be reason to use serial fieldbuses in some instances (hazardous environments, for example). But even that will change in time I think.
One final point which I will have to revisit in more depth: Determinism. This article and others recently have insisted that Ethernet is not deterministic. PROFINET is. For four one-minute explanations see the Performance videos at MinutePROFINET.