I left the country Sat evening and since then the attainable rate has dropped by a third and the speed by almost 40% on downstream and the HEC error rate went pretty high for a while.
A drop in speed of 40% in one go is probably enough for your ISP to be able to call out an Openreach engineer to investigate.
However, I guess it really depends on what the current estimated speed range is, and where your speed now stands relative to that.
I can see on my BQM that the average latency has doubled at 4 AM this morning so I guess that is the line management kicking in a new profile as the SNR went from just over 3db to 6 db
Yes, it probably is DLM.
In the FTTC cabs, DLM leaves the target noise margin at 6dB; it never changes. When DLM intervenes, it normally turns on FEC and interleaving; The former causes you to lose 10-20% of your bandwidth, while the latter causes you to lose 8-10ms of latency.
DLM can intervene further, if previous interventions have not proven effective enough. Then you might lose more latency, and a higher proportion of your speed (I've seen up to around 30% at worst).
Is this normal. I thought in general the line management improved things over the first 10 days not made them worse?
DLM's job is to improve quality, by sacrificing speed and latency.
You should see DLM as having 2 jobs: the first is to establish a stable, quality connection quickly: this it does within 2-3 days since installation. Its second job is to keep monitoring that quality is maintained.
The reason the second part is required is because VDSL2's biggest downside is that it suffers from the effects of crosstalk - noise from other subscribers. Because of this, you can expect the noise levels to only ever increase after the initial installation - randomly - and, hence, expect that error rates increase. This has the dual effect of reducing speed and, eventually, triggering DLM, which then reduces speed further, and adds latency. A double whammy.
There are two solutions:
a) vectoring, a technical solution requiring a BT rollout which might be a few years away.
b) G.INP, a new technique that DLM can use instead of FEC/interleaving, saving that extra drop in speed and additional latency. BT are rolling this out now, so we don't have much experience.
For now, you should expect that DLM will keep watch over your stability, and that you'll lose speed.
Finally can anyone explain very simply what causes HEC errors. The Wikipedia explanation is head bending
I understand the ADSL and VDSL2 protocols fairly well, and thought that HEC errors only applied to a protocol carried by ATM (ie original ADSL only) and PPPoA. The error indicates a problem in the ATM cell addresses.
In FTTC, we use ethernet as the lower protocol layer, and PPPoE; HEC shouldn't apply.
The thing to watch is the ES rate, and see how many accumulate over 24 hours.