If the line looks ok, how was it able to carry more data before ? Does the snr start at 3 and then move up if it detects instability ?
The line looks pretty perfect, very few errors, really steady SNRM, attenuation quite low etc.
The only downside seems to be QLN levels, indicating a general & steady noise issue.
Values around -140 would be typical. However your values are only around -100 (noisier).
My own connection has QLN values closer to -140, but it does exhibit spikes of noise.
I suspect you are asking about SNRM, not SNR values.
Target SNRM is 6.0dB & your connection seems to more or less flatline around 6.0dB.
If SNRM was higher, you would expect to see higher attainable rates, possibly achieving higher sync speeds if diconnecting/reconnecting at that time.
However, as you have no spare margin, your sync speeds & attainable rates wouldn't really change much.
In other words, it looks like your connection is currently achieving the highest speed with good stability that the current conditions will allow.
I can only imagine that your connection may have started off on a wide open speed profile, possibly with higher sync speed than attainable rate due to SNRM dropping to much lower levels (maybe 3dB or less).
Eventually DLM probably lowered speeds which had the effect of raising SNRM back to 6.0dB.
This is pure guesswork & it's a shame that you don't have any stats from the higher speed period for comparison purposes.
Is interleaving currently on ?
No, the values of 1 & 1 shown against D: signify that interleaving is OFF for both DS & US.
Do you hear a constant & steady noise on the phone when running an audible quiet line test that maybe wasn't present when your speeds were higher?
3 days of stats look like a really solid, trouble-free connection.
It is therefore a complete mystery to me that your connection did achieve higher speeds, with the engineer reporting a really high attainable rate.
How far would you say you are from the cabinet to which you are connected?
Your stats suggest around 600m or so.
The engineer's comment actually suggests around 100m or so.