Print

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

[les-70]

   I think I am aware of Zen and Plusnet advice re the FTTC DLM and the general view of it being error seconds and 0.5, 1.0 or 2 per minute being possible thresholds. 

 I therefore looked through all the non interleaved FTTC connections in my dslwebstats to see who had FTTC and what the max ES rates were for non interleaved connections.  i think the winner with most errors but on fast path is BaldEagle1    and his ES rate looks to be a bit over 30/hour.  I am sure he could give an accurate figure.   It looks like no one exceeds much more than about 30-40/hour as an average and if no one is at a higher rate then I assume more than 30-40 will lead to intervention.  This makes me suspicious that that is the actual threshold for most on us.    If we get a big enough sample I think that mydslwebstats is good way of testing reality.

  My failure     was to also check re capped speeds and check the number of interleaved and non interleaved connections.   Maybe another day or someone else!

[Ixel]

I see. Also take into consideration the possibility of the DLM profile not being same for every connection.

I'm not sure if someone has posted it a while ago, but I stumbled into an FTTC training doc, a little old but nonetheless has the same information regarding DLM. http://johnboyle.me.uk/joomlazen/images/resources/tech_info_fibre/FTTC%20Training%20Doc%20(MWright).docx

It mentions what DLM takes into consideration, a table of the MTBE and such like (good and bad thresholds, not just bad thresholds), colour coding that DLM does, and so on. Whether it's accurate, well, don't ask me. So far though I'd say for my own connection it seems reasonably accurate. I'm currently on 49/15 (can easily get 80/20, but lost the sync a few days ago due to multiple re-syncs by the ASUS router on buggy firmware in the early hours of the morning). DLM hasn't relented yet, but I have a hunch that DLM isn't relenting because my error seconds aren't below the 'good threshold' (which I worked out from the MTBE to be 'less than 288 error seconds' I think) for the speed profile. But it might also be that I have a barrier and must wait a bit longer, we'll see. I do have the ECI /r modem, so if it doesn't budge in a week or so then I'll plug that back in and see what happens.

[Chrysalis]

its 2880 I am very sure of it.

Someone on tbb posted they were fast path with 27xx and then interleaved at 28xx.

[les-70]

  I am surprised at that given the mydslwebsats cases.  There are certainly a lot of issues but my feeling is that the highest sustained error rates on not-interleaved and not-capped lines ought to give a good guide to practical values of errors that your not likely to be able to exceed for very many days.

 The biggest complication maybe not knowing which of the three DLM settings apply to people.    That said I am surprised that no one seems to be beating Baldeagle1's 30-40 ES/hour.   No one gets near the 120 corresponding to 2880.

 I am thinking of having a go at beating  Baldeagle1     I get similar or slightly greater values of errors when running without a speed cap.  I suspect the result will be interleaving but it is worth a try.

[Bald_Eagle1]

FWIW, my connection very rarely ran on fastpath when it could achieve higher speeds (i.e. before crosstalk reduced them).

That was undoubtedly due to significantly higher error counts, causing interleaving to be applied.

On the lower speeds & since the October 2013 firmware updates, my connection is now on fastpath for the majority of the time with typically around 900 - 1000 DS Error Seconds, 6000 DS RSUnCorr & 1300 DS CRC errors per day.

Interleaved at around 30 Mbps, I would typically see only around DS 110 Error Seconds & low totals of DS RSUnCorr & DS CRC errors.

[les-70]

    I have just started my un-capped test run with a sync of about 72Mb/s.   I am expecting from recent short tests an ES rate of about 15/hour at quiet times peaking to about 120/hour in the early evening period.   I do not like to guess the 24 hour average error rate as I have not run it like that for long. My best shot is ~40ES/hour.

   I will give it week or until the DLM notices!

[NewtronStar]

A question for BE1 or anyone

Which do you prefer a DS sync @ 22300 Kbps non interleved or DS sync @ 30000 kbps interleaved ?

As you know my downstream has never been moved onto fastpath since day 1 with infininty thats going on 3 years now, I could capp my line to 25000 kbps for a month or two and see if the DS get's moved onto fastpath It's just I don't know what a non interleaved DS line feels like.

[Bald_Eagle1]

As I'm not an online gamer, latency isn't really an issue for me.

Therefore my preference would be 30000 kbps interleaved for faster downloads.

Along with the higher interleaved DS sync speed, I also had a higher US fastpath sync speed of 5700 kbps compared to today's 4300 kbps, which helped when uploading large files to the office when working from home.

[NewtronStar]

As I'm not an online gamer, latency isn't really an issue for me.

Therefore my preference would be 30000 kbps interleaved for faster downloads.

Thanks BE1 my online gaming days are non existence these days but still enjoy an offline game here and there, i'll not be in a hurry to get DS fastpath so i'll stick with what i have 

PS hope you had a good holiday as we are deffo back to long winter nights again 

[Chrysalis]

A question for BE1 or anyone

Which do you prefer a DS sync @ 22300 Kbps non interleved or DS sync @ 30000 kbps interleaved ?

As you know my downstream has never been moved onto fastpath since day 1 with infininty thats going on 3 years now, I could capp my line to 25000 kbps for a month or two and see if the DS get's moved onto fastpath It's just I don't know what a non interleaved DS line feels like.

in your case I would probably prefer the 22.3 but when downloading large games etc. the extra 7mbit can help so I guess isnt clearcut, but it depends how you use the connection really.

Lower latency speeds up web browsing as web browsing is latency sensitive, it will improve the experience of ftp, cli ssh, it will keep gamers happier who like lower latency, it can also help downloads because downloads can be rtt limited.  My experience is that also netflix buffers up to super hd quicker as well.

If you were to ask me would I have 70mbit interleaved or 60 fast path, that is a definite no brainer.  Your question is a bit harder as 30 vs 22 is a bigger %.

It is a shame you cannot choose target snrm on FTTC, then you could choose maybe 9db and have fast path.  We also seen on kitz line recently the extra snrm on fast path was just as effective as interleaving in mitigating her errors.

[Ixel]

A question for BE1 or anyone

Which do you prefer a DS sync @ 22300 Kbps non interleved or DS sync @ 30000 kbps interleaved ?

As you know my downstream has never been moved onto fastpath since day 1 with infininty thats going on 3 years now, I could capp my line to 25000 kbps for a month or two and see if the DS get's moved onto fastpath It's just I don't know what a non interleaved DS line feels like.

in your case I would probably prefer the 22.3 but when downloading large games etc. the extra 7mbit can help so I guess isnt clearcut, but it depends how you use the connection really.

Lower latency speeds up web browsing as web browsing is latency sensitive, it will improve the experience of ftp, cli ssh, it will keep gamers happier who like lower latency, it can also help downloads because downloads can be rtt limited.  My experience is that also netflix buffers up to super hd quicker as well.

If you were to ask me would I have 70mbit interleaved or 60 fast path, that is a definite no brainer.  Your question is a bit harder as 30 vs 22 is a bigger %.

It is a shame you cannot choose target snrm on FTTC, then you could choose maybe 9db and have fast path.  We also seen on kitz line recently the extra snrm on fast path was just as effective as interleaving in mitigating her errors.

Interestingly you can choose target SNRM on the ASUS DSL-AC68U, I've tried it and it works. I'm trying to get my capped sync profile lifted, but I think my ES isn't below the 'good' threshold so DLM currently won't increase the profile after it was decreased some days ago from 74/20 to 49/15 due to previous version of ASUS firmware causing multiple re-syncs in the early hours. Either that or I'm once again stuck on 49/15 for a year or more. So frustrating that ISP's don't have the ability to reset DLM years later from Infinity being rolled out.

[Chrysalis]

yeah if I understand that info right then ES needs to be below 288 to get a DLM reversal?

[les-70]

   Looking at the collection of lines on mydslwebstats I think there are many FTTC lines that seem stuck on their current banded or interleaved states with ES quite a bit less than 288.  Again the reality depicted by lines on mydslwebstats does not seem to match the 2880 theory.     If your on an band most should be in that band  e.g. 288-2880 for the fast band.

  The expected ranges according to the doc are either 72-720, 144-1440, or the 288 to 2880. mydslwebstats suggest not many or any are on the 288-2880 band.   Zen may be an exception I did not look who might be on Zen.  I suspect most may be on 144-1440.

[Chrysalis]

well plusnet claimed earlier in the year when asked people are on the speed profile, would they lie, or have they misjudged who knows?

I estimated kitz last week to have less than 2880ES when she got interleaved but unfortunately she never confirmed that 24H stats from her router.

I think its not 100% accurate to guess daily ES from those graphs tho.

I think people only get unbanded if they syncing at the max cap of the band.  I am more referring to interleaving levels.

[Ixel]

well plusnet claimed earlier in the year when asked people are on the speed profile, would they lie, or have they misjudged who knows?

I estimated kitz last week to have less than 2880ES when she got interleaved but unfortunately she never confirmed that 24H stats from her router.

I think its not 100% accurate to guess daily ES from those graphs tho.

I think people only get unbanded if they syncing at the max cap of the band.  I am more referring to interleaving levels.

It's all confusing, as I expect there's more to the DLM algorithm than just those ranges. The training doc states that it also takes into account the history of the line, the FEC count, and variations in things such as the rate (presumably either attainable rate or sync rate, not sure), that's assuming it's all accurate information.

Unfortunately the unbanding didn't work that way for me. More than a year ago I was stuck on 60/20 although I had an attainable rate of over 90 meg down. Around late December 2013 to January 2014 I was back at 80/20. I wasn't interleaved on either band. Suffice to say it took several months before I managed to get 80/20 back, either due to a reset that was performed at the cabinet perhaps (firmware upgrade rolling out at the time for the HG612?), or something relating to DLM's algorithm decided to let me have 80/20 again.

As it stands, at the moment I'm officially below the estimates on both clean and impacted ranges. Attainables are currently 98 meg down and 21 meg up (reduced transmit power which increased the downstream attainable rate but reduced the upstream attainable rate, obviously relating to the attempt to cut down the number of ES on my connection - though ASUS doesn't report an accurate number I'm trying to count it by other means). I may return to the ECI /r if it doesn't relent in a couple of days. Another idea I had was what happened if interleaving was put on my connection, perhaps after a few weeks DLM might then unband and remove interleaving at the same time (as my connection ES should be within '< 288').