Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[NewtronStar]

I find these higher ES threshold values way to high before the DLM kicks in, as this evening got 60 errored seconds in under 1 minute due to faulty table light switch it was fizzling & the light flickered the HG612 has done a re-sync very close to this event with a retrain 6 code.

You can have a look at MDWS and see the results for yourselves

[ardsar]

Whatever DLM is now running it seems very slow at reacting.

On the 20th November it nudged me up from approx. 40/14 to 47/17 having dropped the ASUS DSL-AC68U completely a few days prior. Since then no change despite what look like very good stats and just one reboot to change some dsl params...

Total time = 7 days 7 hours 44 min 16 sec
FEC:            6618766         250663
CRC:            119             55
ES:             20              22
SES:            0               13
UAS:            33              33
LOS:            0               0
LOF:            0               0
LOM:            0               0

Either DLM isn't running nationwide just yet or it's forgotten about me 

Im in a similar situation. Think dlm is punishing those that used the asus dsl-ac68.

Sent from my LG-D855 using Tapatalk

[Ixel]

Whatever DLM is now running it seems very slow at reacting.

On the 20th November it nudged me up from approx. 40/14 to 47/17 having dropped the ASUS DSL-AC68U completely a few days prior. Since then no change despite what look like very good stats and just one reboot to change some dsl params...

Total time = 7 days 7 hours 44 min 16 sec
FEC:            6618766         250663
CRC:            119             55
ES:             20              22
SES:            0               13
UAS:            33              33
LOS:            0               0
LOF:            0               0
LOM:            0               0

Either DLM isn't running nationwide just yet or it's forgotten about me 

Im in a similar situation. Think dlm is punishing those that used the asus dsl-ac68.

Sent from my LG-D855 using Tapatalk

I've always found that DLM is slow to react to improving the downstream on any device, though is quicker to react to improving the upstream (if it needs to).

[simoncraddock]

Looks like DLM finally took a sniff at my line this morning around 9.30am (odd time).
Increased from 47/17 to 54/17, so it looks like the affects of the ASUS are slowly being undone.

[les-70]

Pleased to hear that.  My limited understanding of the DLM is that if you had for example 3 successive stabilizing actions taken by the DLM that the last one will come off quite quickly if the errors go low enough, but each of the other hits will take an increasingly long time of good performance to convince the DLM to reverse things.  So if you have another level to remove that may need an even longer wait.

[xreyuk]

It seems as though after reading this thread I'm none the wiser on DLM!

I spent 2 months with a manually banded line at 38,000Kbps with an attainable of 51,000Kbps and an SNRM of 9.4dB. Error rates were lower than 2 ES per hour, and even after 2 months my connection was not put on fast path.

It had only previously been disconnected 3 times in 3 months (once manually, twice by DLM which was in the first two days of install) and it still didn't recover.

I've let it go back to whatever it wants and I'm currently sat at 44,000Kbps and an SNRM of 6.4dB (I was tired of slower downloads), ES are currently at 8 ES/hr after 20 days uptime, and again no change.

Would we expect me to be put on fast path based on what we've found?

I'd love to get on fast path because I'm a big gamer.

[marjohn56]

Who knows... Only OR and they won't tell anyone, maybe a freedom of information request by your MP may help!

Whatever it is, it seems it's an exponential curve when it comes to restoration.

[les-70]

  My previous analysis of 10 days of all the MDWS interleaved connections and a few transitions suggested that you needed to be below ~20-40ES/day, often for a very long while, to go to fast path, an exact figure just can't be deduced.  Most MDWS interleaved connections get 30-200ES/day and over the length of record that I looked at seem stuck interleaved.

  @xreyuk  Interleaving gives a very big CRC/ES reduction so recovery needs low values.  Where your correct to be puzzled is that there are many interleaved people who seem stuck interleaved whose long term daily ES rate is significantly less than suggested values of 60 standard and 120 fast. Line history should influence the ease of recovery but to many it does look its line archaeology that seems to count.  As noted above a value more like a sustained less than 30ES/day seems a better fit to experiences of a return to fast path.

[kitz]

there are many interleaved people who seem stuck interleaved whose long term daily ES rate is significantly less than suggested values of 60 standard and 120 fast.

The 60/600 profile which I posted last week dates back to 2009.   

I recently came across a new algorithm dated Jun 2012 which is labelled NGA, its worded strangely but if you do the maths then it comes out with exactly the same figures as quoted by Zen. Ive contacted someone for confirmation if theres anything later but so far they are keeping stum. I cannot find anything that specifically relates to FTTC that is newer than the 30/300 profile.

As mentioned previously there is another algorithm which decides how soon your line recovers, and it certainly implies that recovery is different depending on whether the DLM kicked in due to MTBE or MTBR.

[kitz]

PS - dont forget part of that step... and as previously mentioned... from what Ive been reading in the court case about the flow chart..  seems to suggest that for those who have gone deeper down the DLM rabbit hole, then their SNRm is monitored for 14 days before positive changes are made..  and if your SNRm varies by more than say 4dB over that 14 day period then you are stuffed

[xreyuk]

  My previous analysis of 10 days of all the MDWS interleaved connections and a few transitions suggested that you needed to be below ~20-40ES/day, often for a very long while, to go to fast path, an exact figure just can't be deduced.  Most MDWS interleaved connections get 30-200ES/day and over the length of record that I looked at seem stuck interleaved.

  @xreyuk  Interleaving gives a very big CRC/ES reduction so recovery needs low values.  Where your correct to be puzzled is that there are many interleaved people who seem stuck interleaved whose long term daily ES rate is significantly less than suggested values of 60 standard and 120 fast. Line history should influence the ease of recovery but to many it does look its line archaeology that seems to count.  As noted above a value more like a sustained less than 30ES/day seems a better fit to experiences of a return to fast path.

Thanks Les, I think I just got tired of waiting for it to restore and let it do what it wants. I'm still getting around 21ms to the BBC, so it's not like I'm sky high for gaming, just would have been nice if it was lower!

[Chrysalis]

PS - dont forget part of that step... and as previously mentioned... from what Ive been reading in the court case about the flow chart..  seems to suggest that for those who have gone deeper down the DLM rabbit hole, then their SNRm is monitored for 14 days before positive changes are made..  and if your SNRm varies by more than say 4dB over that 14 day period then you are stuffed

so the recovery process is now changed since the court case?

[les-70]

  Tbailey2 has been over 1440 ES/day since the 5th December.  That seems to be the first solid evidence in MDWS stats that the Plusnet ES limit is 2880/day. 

[simoncraddock]

DLM adjusted my line again this morning around 9.30am, up from 54/17 to 54/19, 5 days after the last alteration. It would seem that DLM is a good deal quicker at raising the Up Stream than Down stream. Fingers crossed the next adjustment should restore fast-path on the downstream.

[kitz]

PS - dont forget part of that step... and as previously mentioned... from what Ive been reading in the court case about the flow chart..  seems to suggest that for those who have gone deeper down the DLM rabbit hole, then their SNRm is monitored for 14 days before positive changes are made..  and if your SNRm varies by more than say 4dB over that 14 day period then you are stuffed

so the recovery process is now changed since the court case?

No reading the court case notes, led me on a path to searching for something else which I found more info on.

Too long to explain atm, but I think I posted a link a few pages back to a BT patent which showed the algorithm.   Lots of reading to do to understand it, but I will at one point attempt to summarise it all.