Because of his interleaving depth, which is:
D: 1435 1
Which is quite a high level, so something must be causing errors on the line or the modems been switched off a few too many times in quick succession.
Knowing the depth of interleaving alone is insufficient; the interleaving works using a two-dimension array, and the depth only shows one dimension of the array - you need the other dimension to determine the "amount" of interleaving.
In this case it is the value
I 69 255
So interleaving uses an array 1435 x 69.
However, I find that even knowing this isn't particularly useful. By far the best measure of interleaving is to look at the parameters set by DLM directly, and particularly the "delay" parameter:
INP: 3.00 0.00
delay: 8.00 0.00
OP:
An INP of 3 symbols and a "delay" of 8ms is the standard "first attempt" setting for DLM to choose. That implies that the error stats are bad enough to warrant intervention (or were, at the time DLM intervened), but not so bad that DLM has had to perform any additional intervention.
I do find it useful to look at the parameters for FEC, as these vary quite considerably. In this case, the values are:
R: 14 16
N: 69 255
This means that the modem is using 14 bytes for FEC protection in every block of 69 bytes downstream; this amounts to 20% overhead. That figure is in line with "standard" overhead for intervention: I've seen values range from 15% to 30%.
Note that the modems have negotiated some upstream FEC overhead (16/255 = 6%), even though DLM has not required it (upstream INP was 0), but interleaving is turned off upstream. This seems to be a regular choice by the modems when there is plenty of spare capacity available upstream.
Back to downstream; when looking at the overall line stats, we see:
Max: Upstream rate = 28094 Kbps, Downstream rate = 96416 Kbps
Path: 0, Upstream rate = 20000 Kbps, Downstream rate = 80000 Kbps
...
SNR Margin(dB): 10.5 10.0 9.7 N/A 6.6 6.5 6.6
The SNR values (just above 6dB) suggest your modem can only just manage to achieve the 80Mbps sync speed alongside the 20% of overhead for FEC; that would match the 96Mbps of the attainable.
Note: In the past, we have found that the Max attainable speed seen when FEC/interleaving is turned on is too high. If DLM removes intervention, it is likely that the attainable speed will reduce to around 88Mbps.
What distance are you from the cabinet (line distance, rather than as-the-crow-flies distance)?
So, what caused the DLM intervention?
On FTTC, DLM is not very prone to intervening because of modem disconnections; I've seen a figure of 10 disconnections in a day. VDSL2 has a mechanism for the modem to report a loss of power to the DSLAM, so it is always better to power-cycle the modem before disconnecting it from the phone line.
The usual DLM intervention comes from monitoring the error rate - in particular we believe it monitors the value that ES changes over a 24 hour period before intervening, or increasing intervention. It may then monitor the FEC counters to aid its decision on de-intervening; this aspect is less clear.
Right now, your stats show:
OHF: 129019251 286885
OHFErr: 602 2
RS: 3296679127 3471720
RSCorr: 37382999 43
RSUnCorr: 61068 0
and
AS: 135975
That shows that the FEC process has corrected 37,382,999 errors in 135,975 seconds of uptime. That's 275 errors per second, or 1 million per hour over a 37 hour period.
That sounds like a lot - it means that around 1 in 88 blocks of data needs correcting, so it is no wonder that DLM has intervened - and I think it is likely that the intervention will stay.
However, it also looks to be a successful intervention: The FEC process has only failed to correct 61,068 blocks in that time, which has led to only 602 CRC errors; even better, those errors have occurred over only 122 seconds (the ES value): so less than 100 ES's per day. It is unlikely that DLM will need to add more intervention.
For comparison, my line (with almost identical attenuation) has no DLM intervention. It runs at around 5,000 CRC's per day and 500 ES's per day.
Pages in this thread: 
Print Thread
deleted
