Hi all,
I'm curently spending a lot of time trying to put hard numbers on the costs choosing a protocol produce. These costs are not only the license price, but there are also performance costs on the PLC itself and bandwidth costs on the network.
I did my best to white up a detailed analysis for the Beckhoff ecosystem
Hope this might be interesting for the one or the other.
Chris
Very interesting indeed. As with your recent Siemens article, I am again astonished that a PLC's periodic task is affected at all by communications load. In Rockwell, comms responses suffer from an overload condition, not the PLC scheduled logic.
Well in the Beckhoff article I guess it's similar (if you use the native protocol) ... ADS is almost not measurable at whatever rate I read. OPC-UA is what brings the device to it's knees, if the device has a lot of work to do.
I like to compare it with languages ... For me OPC-UA is somewhat like the proposal of everyone learning a common language "Esperanto". If we did that we could all speak with each other. However not speaking in your native tounge costs mental power. Speak with 100s of people at a fair in a language that's not your native one wears you out.
If however someone is able to speak in multiple languages as if they are his native language, communication becomes easy for everyone.
That's what I'm trying to do with our ToddySoft Connect driver suite.
Hopefully nobody is dumb enough to poll via the Read service over OPC UA when subscriptions are available 
Well ... let's say on my bench here ... OPC-UA subscriptions really seems where the different implementations separate widely.
My bench was literally doing benchmarks 24h a day for the last week on all of my hardware using all of the protocols I had and all the modes I could use them in ... I'm currently working on writing the "cross protocol comparison".
And ... I think on my S7 devices I couldn't get more than 2000 tags as that seems to be a limit of the OPC-UA server/license ... it won't let me compile if I add more.