I am getting the usual cyclic upstream line 3 behaviour, 6dB high time, 1-1.5 dB low time and the upstream target is 6dB. I’ve just noticed something. The line dropped for some reason or other, but anyway, it dropped at such a point that the upstream SNRM vales were as stated, with a really low low. And that mean that the upstream rate was really good, at 499 kbps, which is about ~100k better than it usually is.
This time I did some speedtestests, with speedtest2.aa.net.uk and the ookla speed tester, because I wanted to find out what the combined upstream rate was like when all four lines were roughly the same rate upstream, all at or just over 500kbps u/s. To my surprise, I found that the upstream combined rate results were rubbish compared to what I have been getting recently. speedtest2.aa.net.uk was ~1.30Mbps with the 499k upstream line 3 rate and the expected rate was 1.5-1.6Mbps. So as an experiment, I reset the line 3 upstream SNRM so that it was back at the target of 6dB and the upstream rate then went back down to 389k
Live sync rates:
#1: down 2807 kbps, up 519 kbps
#2: down 2700 kbps, up 544 kbps
#3: down 2964 kbps, up 389 kbps 6db upstream SNRM, 6dB upstream target
#4: down 2553 kbps, up 515 kbps
Now at this unequal upstream rate, when you would expect combined upstream to be bad, the result was the reverse: a combined upstream rate that was a big improvement - combined u/s rate =1.65Mbps at an u/s line 3 rate of 389k.
So in a nutshell: upstream equal speeds - bad upstream combined effective performance
Note: The line 3 downstream rate is about 30kbps lower, although I can’t see how that could affect things too much, unless it’s something to do with too many errors slowing things down?
So how can the upstream combined rate be 300k worse when the rates are equal ? All the other upstream rates are unchanged, btw, so that isn’t it.