Ejs, I wasn't believing the numbers that I was seeing, don't misunderstand me. :-) It seemed to me that there was some madness in the modem ui. I didn't know that about not covering the whole cell, you're absolutely right, that makes a world of difference and if these numbers were the truth then I would have a huge problem which would be noticeable, just as you say. I completely agree with you on this.
I also have not got as far as re-reading stuff on the basics of ATM and looking again at certain parts of the G.992.3 specs.
Many thanks for putting me straight.
@Burakkucat - I shall take your systematic check method on board. There are some other asymmetries that are unexplained - the max latency reported by clueless during heavy traffic is vastly higher ~100% greater - for line/modem 1 compared to the other lines. That is true for different modem models too, before and after a swap out. I wonder if the Firebrick is thrashing that link harder than the others. Line 1 happens to have the fastest upstream sync so it has the highest speed value set in the associated speed shaper object in the Brick. This was discussed elsewhere. It's just that there do seem to be various things that make link 1 different in certain odd little ways.
