Developments to watch: Will one super network end Ethernet wars?
Could the IEEE Time Sensitive Networking (TSN) effort unify industrial networks and fractured Ethernet communications? If it succeeds, it could boost productivity significantly to refocus time, effort, and resources on more productive things.
First it was fieldbus wars, then a proliferation of device networks, safety networks, motion networks, and many Ethernet protocols across and within industries. Many are tired of the hassles and inefficiencies of so many communication protocols. What if everyone accepted one high-speed, wide-bandwidth protocol that performed better than any of these? One communications standard? "Ha!" you say; "It's past April Fool's day."
An IEEE task group is working on a new Ethernet protocol (IEEE 802.1) to support control systems called Time Sensitive Networking (TSN); Eddie Lee and Todd Desso at Moxa gave some initial observations to CFE Media recently. The Ethernet protocol would work on the Ethernet physical layer, reliability is in the seven nines (99.99999%), it offers plug-and-play time synchronization with typically less than a microsecond of jitter, and major end users and vendors are involved.
Lee and Desso cited a recent ODVA annual meeting presentation, noting that the IEEE AVB Task Group has completed numerous projects contributing to new standards for improved time synchronization and reduced probability of packet loss even in converged networks. However, more work is needed by the newly named 802.1 TSN Task Group to build upon the existing standards. New initiatives include:
- Making IEEE 1588 Precision Time Protocol simpler to use, more robust, and more accurate, while keeping its plug-and-play capabilities.
- Offering the ability to reserve bandwidth for particular streams, and to guarantee those streams a low, specified latency and zero congestion loss.
- Allowing a single data stream to take multiple pinned-down paths through the network to ensure that equipment failure will not cause packet loss.
- Allowing any data stream to reserve this Quality of Service (QoS), whether unicast or multicast, regardless of high-layer protocol, whether layer 2 or layer 3 end-to-end.
- Allowing convergence of all the usual existing QoS into the TSN network (at reduced bandwidth) while maintaining the TSN guarantees.
Due to the multi-faceted nature of technology standardization and collaboration, many subject matter experts from leading companies may have varied interpretations for the best path to conformance. Some appear to be forging ahead in bringing a TSN solution to market prior to formal ratification, Lee and Desso observed.
I would strongly encourage various collaborators to work together to make one version of TSN. The process control world tried to make one fieldbus. (That didn't work; the "standard" contained seven versions.) Industrial device and sensor network designers had an opportunity to learn from that example and didn't. Then industrial Ethernet divided into how many protocols? The industrial world faces many complexities: TSN shouldn't be one of them. In a few years, when IEEE is finished, I hope we have one version, rather than another nonstandard standard.
Benefits? What if every rail line were a different gauge, or every outlet carried a different voltage, or every smartphone brand used a different wireless protocol? Perhaps now is the time to create real value in networking instead of creating obstacles that limit the ability to integrate, plug and play, and exchange information.
- Mark T. Hoske, content manager, Control Engineering, CFE Media, firstname.lastname@example.org.
IEEE has more has more information.
Comment, May 16: "Read with some interest your squib on page 36 of the latest Control Engineering. [May 2014, posted above with more details] Maybe Ethernet wars will go on forever. Let's hear it for industrial competition and free enterprise. Cheers!"
-Dick Johnson, Control Engineering senior editor, retired