glass scales via ethercat?

If you're wanting to use the scales as part of a closed loop on a machine with typical travel speeds, then ethercat isn't going to be fast enough.

To close a loop, you need cycle times measured in microseconds, whereas ethercat cycle times are measured in milliseconds.
RS485/422 is even slower, so it all depends on what you're actually wanting to achieve. If you're just wanting to measure position, then either method should work, but if you want run closed loop, it's going to be too slow.

You can replace the encoder on the servo motor with a glass scale and let the servo drive do its job. That would typically mean an incremental differential quadrature signal which is what most cheap glass scales have. Maybe you can get the encoder position from the servo drive over ethercat and save the need for a separate encoder.
Ethercat works fine with closed loop in linuxcnc, the servo loop time is 1 millisecond or thereabout.

Edit: this was already tested in an earlier post.
I can add that I searched around out of curiosity and found a few rotary encoders that send absolute position with timed 1 ms intervals  or slower over rs485. Then that communication should be done by an ethercat unit, decode the value and make that available in a pdo. Does such an ethercat unit exist?

For scales you might be able to use one of these buit not much cheaper than Beckhoff
www.icpdas.com/en/product/ECAT-2092T
There are also ethercat specific encoders
www.encoder.com/ethercat

To clarify: I was thinking about using RS482/422 via Ethercat Beckhoff modules. I use RS485 for my VFD via USB as well.

I don't disagree with the statement regarding shorter cycle times. However, I am not so sure about how linuxcnc processes this. It has been my understanding that closed loop in linux cnc is part of the servo thread and thus usually defaults to 1 ms - faster is certainly possible, but becomes more difficult. Furthermore, ethercat with cycle times below 1 ms is also asking for trouble.

What I do not know is how much of a time offset results from using ethercat to input glass scales. I would image that between t0 (encoder position is read) and the processing of the information within the servo thread a certain time (number of cycles passes). I do not know how many.