Kitz Forum
Broadband Related => Broadband Technology => Topic started by: sevenlayermuddle on August 05, 2009, 09:35:33 AM
-
As is well documented for anybody with time to read my posts, I've now been waiting nearly a year for DLM to reduce my target margin from 15dB.
We had a power cut earlier in the week. When the power was restored I got a low sync as I always do following a power cut, presumably because everything electrical in my own and neighbours houses is all springing to life at exactly the same time. And also, of course, because my router loses it's SNR tweak. I know, installing DGTeam f/w would help, but for various reasons I've not done so.
So, I'm currently synced at 1920 with a 18dB margin, vs my normal >4000 @9dB. In just over 48 hours (175,000 AS), I've had a total of 57 ES, 74 HEC and 39 CRCs.
I'm not planning any activity that needs high speed access, so I'm going to leave it as it is for a few weeks, and will report back whether DLM then reduces my margin. I'm predicting that it won't, but will be happy to eat my words if it does. :-\
-
There does seem to be a lottery aspect to this process. As I recorded several months ago, my target noise margin did come down from 15 dB to 6 dB in three fortnightly steps, but it didn't last long. Leaving it untweaked as you're doing is obviously giving DLM the best opportunity to work, and it will be interesting to see how it goes.
-
>> with a 18dB margin
:no:
>> at 1920 with a 18dB margin, vs my normal >4000 @9dB
fits very nicely that on average each 3dB is costs approx 800kbps of speed.
- not good for you though :(
>> but will be happy to eat my words if it does.
Be nice if it does.. but its never seem to have shifted before for you.
Would be good if the power cut actually kick started the DLM to work how the theory says.
-
My theory is that in return for a low sync speed I will get very low error rates, and that's what may convince DLM my line's not so bad.
But wouldn't it be nice if BTw would tell us more about how DLM actually works?
-
18dB is one hell of a margin!
I had a "strange occurence" on my line yesterday (ADSL2+ with Be) resulting in no or minimal throughput despite an apparently normal (around 20mbps) sync. After a lengthy call to support, who could see lots of errors, my target was raised to 9dB and interleaving re-enabled (I'd been on 3dB Fastpath for over a year with no problems) so I'm having to "live with" only 16mbps :angel:.
The idea is to run for 24 hours at 9dB then revert to standard if the error rates are OK. After 15 hours of that I have 0 (zero) absolutely NO ERRORS of any description whatsoever using a DG834GT with DGTeam firmware!
I can only assume there was some temporary interference that has now gone, but if it re-appears I'll know what to do, and that it can be done quickly!
The contrast with the BT system is enlightening.
Firstly, the willingness & ability of Be support to check, investigate and adjust in real time - they rang me back on my mobile to facilitate line tests. Really helpful.
Secondly, the flexibility of NOT being a slave to BT's DLM systems. :P
My Netgear will usually hang on reliably down to about 1dB margin before giving up.
-
The contrast with the BT system is enlightening.
Firstly, the willingness & ability of Be support to check, investigate and adjust in real time - they rang me back on my mobile to facilitate line tests. Really helpful.
Secondly, the flexibility of NOT being a slave to BT's DLM systems. :P
Sounds good to me, I think I'm envious.
Actually, I'd like to think that BTw might come under political pressure to overhaul the DLM algorithms in pursuit of the goal of getting everybody in the country up to a sensible minimum speed.
Of course, there are cirumstances in which DLM will want to adjust the target margin to stabilise a line, but the existing DLM does seem a little inclined towards a sledgehammer approach of 'raise it forever, at the slightest excuse'.
-
Actually, I'd like to think that BTw might come under political pressure to overhaul the DLM algorithms in pursuit of the goal of getting everybody in the country up to a sensible minimum speed.
Perhaps some sort of more advanced learn/adapt fuzzy logic systems or neural network, but anything would be a truly enormous task.
-
from orainsear -> 'but anything would be a truly enormous task'
But SOMETHING needs to be done about the way it works at the moment. The phrase 'using a sledgehammer to crack a nut' sum it up nicely I think.
TTFN
toulouse
-
From what I understand of it, one weakness is that I think it uses the FTR as one parameter in the decision to raise the target. Basically, in my understanding, it won't raise the target if doing so would breach FTR - and that should operate as a safety valve to prevent a long line that's capable of, say, a ssteady 2mbps at a 6dB marging being reduced tenfold to 0.2mbps with a 15dB margin (assuming 600kbps for each 3dB).
However, FTR isn't ever re-calculated after the initial training period, despite the fact that the initial FTR becomes innapropriate after improvements, such as filterred sockets, better routers, etc. I do strongly suspect that's part of my own specific problem, since my typical speeds have improved at least three-fold since my iniital training, thanks mainly to chopping off the bell wire and getting a better router.
As always, I have to stress that's only my understanding, based on what I've dug up form my own research. I could be wrong.
-
Hi
Quote My theory is that in return for a low sync speed I will get very low error rates, and that's what may convince DLM my line's not so bad.
I wouldn't hold my breath.I had some problems 3 o4 months ago .Some contractor part chopped through the undergroung cable . I ended up with 15db target SNR . There were a few hundred errors during that particular incident.Since then My errors have been very low .
Total Up Times (From SF counts):
WAN: 10 days, 20:41:17
LAN: 11 days, 03:53:08
CRC: 12
LOS: 0
LOF: 0
ES : 12
Noise Margin: 11.9 dB
Connection Rate: 6592 Kbps
Line Attenuation: 31.0 dB
Power: 19.8 dBm
Max Rate: 7392 Kbps
The above are for the last 11 days . I rebooted then when my snr margin was at its highest to get slightly better synch speed.
These figures are typical for the last 3 to 4 months .At the moment I am tweaked to 12db.
Similar to Roseway ,before the last problem my Target had come down in stages to 9db from 15. in about 6 weeks.Now no sign of movement.
Regards Jeff
-
:'( :'(
Thanks for that Jeff, in view of these comments I'm thinking I'm wasting my time.
But I'll give it a few weeks anyway, or at least until the next time I need to download something big, like a DVD .iso image. For the things that I do, that's the only time dismal speed really hurts much.
The next thing I'm tempted to try is to re-instate my bell-wire, and reconnect my crappy old router, which I suspect would take my resultant speed down close to, or maybe below, my FTR. But (a) I don't know if I'm willing to make that sacrifice, and (b) it would be hard to conclude anything from it unless I give low error-rates a chance first. There's also a (c) that if I do go below FTR, I might upset DLM even more.
-
...I might upset DLM even more
Well, we wouldn't want that to happen. She (DLM) is a bit a bit of temperamental bitch isn't she ?
-
quote She
you're living dangerously :lol:
Regards Jeff
-
The (only) reason I used SHE, is because there is a lady in my life who, very much like the DLM system is virtually totally unpredictable, I just NEVER know what she's gonna come out with next
Apologies to any of the nice ladies on here who may become offended by such remarks.
TTFN
toulouse
-
>> Basically, in my understanding, it won't raise the target if doing so would breach FTR - and that should operate as a safety valve to prevent a long line that's capable of,
From the document you found a while back.. that was my understanding too.
-
Well well,
Guess what, I never thought I'd see the day, but I just noticed DLM's given me back 3dB, I'm back on a 12dB margin! :thumbs:
I don't know when that happened, but it was still 15dB when I looked a couple of days ago. It does look like just over two weeks with an actual SNRM of about 18-20dB (thanks to a lousy initial sync), with a corresponding error rate less than 20 ES per day on avreage, has done the trick for DLM.
I'm predicting that it won't, but will be happy to eat my words if it does. :-\
Munch, munch :-[
ALong with my 12dB margin, I'm back at 3552 kbps, which with my 56dB line means several (say, 5-10) ES per hour. So I'm guessing DLM won't give me any more, but I'm going to leave it alone (untweaked) for another two or three weeks. If nothing gives, then I think I'll try another experiment, which will be to 'adversely' tweak my SNR to a higher level than DLM's target, and thus get my error rates right down again.
-
It will be very interesting to see if it does go further. Some time ago I went through the same process and it went all the way back to 6 dB eventually, but my ES levels were very low.
-
It will be very interesting to see if it does go further. Some time ago I went through the same process and it went all the way back to 6 dB eventually, but my ES levels were very low.
Well I'll definitely let you know. I'd actually quite like to be able to remove my 'SNR tweak' for good, but I'd want a 9dB target for that.
I'll also be intersted to see whether my hard-won DLM improvement is any more permanent than the improvement brought about by BTw's manual override, following a request from my ISP. That 'improvement' evaporated after 48 hours, reverting back to 15dB again.
One thing I'm tempted to deduce from my own, yours, and some others experience is that, if the line's stable enough, then 14 days does seem to a critical period over which DLM samples the line before reacting. Mine took about 17 days, but I recall a big thunderstorm at or about day 3, which possibly caused a brief increase in error rates that restarted the 14 day clock.
-
Yes, my margin went down at about 15-day intervals.
-
>> but I just noticed DLM's given me back 3dB, I'm back on a 12dB margin
OMG.. I thought 7LM was going to be stuck at 15dB for eva and a day. Congrats on some improvement at least.
Like you say it will be interesting to see if there are any further developments.
Please do keep us informed.
-
Actually Kitz, it just dawned on me that it's been about year. In fact, driving home I began counting, and thought it might have be exactly 365 days.. now that would have been a dramatically interesting coincidence! In fact, checking back over my emails to ISP, it was about 2 weeks short of the year, so no real coincidence at all.
One curiousity, I have noticed a couple of unexplained spontaneous resyncs over the past few days, following that margin drop. I strongly suspect they were exchange-initiated as my line hasn't lost sync in months, even at much higher speeds than I have now. But I don't run anything like routerstats (shame on me) :-[ so I've no real evidence of it. I almost wondered if my ten-day training period might be getting re-run but, presumably, if that were the case I'd have gone back to 6dB initially, not 12dB?
-
>> if that were the case I'd have gone back to 6dB initially, not 12dB?
I would have thought so.
-
To recap the last few weeks... One day in early August, following a power cut, my router synced at the default target margin of 15dB but, as often seems to happen after a power cut, the sync speed was much lower than I'd normally get at 15dB. Correspondingly, once the the noise had settled, the actual SNR soon rose to about 18 dB. With such a high margin, my error rates became very low, averaging no more than 20 ES per day.
After a littler over 14 days with these low errors, DLM rewarded me with a 3dB target reduction to 12dB.
That 3dB reduction was on 22 Aug. Since then, with a 12dB margin, my error rates have been not bad, averaging about 150 per day. But that's not been good enough for DLM as I've had no additional reduction in target, despite having let it maintain continuous sync for the last 19 days in case it helped.
So, I've now bravely tweaked my margin again, 'adslctl configure --snr199', which has got me an 18.5dB margin, and obviously a loss in sync speed. But I suspect there will also be a big drop in errors.
It'll be interesting to see, in a little over two weeks time, whether DLM has responded favourably to the lower error rates. ;)
One thing I've already noted is that the earlier reduction has stayed with me. That contrasts with the manual reduction which I persuaded by ISP to get done year ago, and which evaporated 48 hours later. I've seen similar accounts elsewhere of manual overrides quickly reverting. I do wonder whether BTw's DLM interface actually enforces a 48 hour timeout on manual overrides, or at least there may be a tick box for it somewhere on the GUI, that the operators either ignore or don't understand?
edited to get the dates accurate
-
You're certainly a rare beast, 7LM, to tweak your margin up. :) But it's almost certainly the right thing to do, to try to get DLM to lower the target further. If it does get to 9 dB, I think I would probably leave it there, because it looks as though your error rate increases quite rapidly with lower noise margins.
-
If it does get to 9 dB, I think I would probably leave it there, because it looks as though your error rate increases quite rapidly with lower noise margins.
Yes, I totally agree, 9dB is probably right for my line, and when it was at 15dB target that's as far as I ever tweaked it. In fact most of my neighbours' seem to have settled at 9, and my line was happy (and so was I) at a DLM target of 9dB for as long as I can remember, before I stupidly upset DLM by too many deliberate resyncs one afternoon whilst experimenting with wiring changes. I really just want to get back to where I was before that happened, without future tweaking.
As an aside,when I tried 'adslctl configure --snr 200' it didn't seem to sync at the first attempt, so I gave up after a few seconds. Rather than keep trying, and maybe trigger alarms in DLM, I then tried '--snr 199' and that worked. I don't know whether '--snr 200' would have worked if I'd tried again, or waited longer. As you infer, it's probably not something people do often. The router's a DG834GT with standard f/w, so I believe 200 should have worked. :-\
- 7LM
-
The router's a DG834GT with standard f/w, so I believe 200 should have worked.
Yes, I would have expected it to work, although it's not something I've ever tried. I use a DG834GT with the DGTeam firmware (which gives a slider control in the web interface to make this tweak) and the slider permits values up to 300.
-
Some time this afternoon, my router resynced itself at 15dB. So I've just removed my +6dB tweak (with a power-off reboot), and whad'ya know, my target's back at 9dB.
The interesting thing about this is that it's only taken a week of that high margin, whereas I think we've concluded from my past experience and other previous reports that 14 days seems to be DLM's normal 'proving' time. I deduce from that that the error rates have been averaged over 14 days, and the low errors of the past 7 days averaged with the slightly higher errors of the previous 7 days, were enough to satisfy it.
Also, there were a couple manual resyncs 7 days ago when I set the +6dB tweak, but that doesn't seem to have spoiled anything. My error rates over the past week have been around 50-70 ES per day, against about 150 ES per day for previous weeks. So it may be the case that DLM looks for a rate between these two before reducing the margin. I guess DLM may also have been influenced by the fact the margin reported by my router was consistently very high compared to the DLM configured target.
What a shame BTw won't just tell us exactly what DLM looks for before reducing the margin, so we wouldn't have to guess. But regardless of DLM's secrets, I feel that I've finally tricked it into reducing my margin when it clearly had no natural inclination to do so. Tonight, 7LM's a happy bunny, and feeling somewhat chuffed. :lol: :thumbs:
-
That's very interesting 7LM. :thumbs:
-
Hi
The mysteries of the DLM my error rates are < 3 ES /day yet it takes Much longer than 14 days to drop my target SNR by 3 db it has taken up to 2 months . still waiting for it to give me back another 3 15 days and counting ::)
Regards Jeff
-
Hi
The mysteries of the DLM my error rates are < 3 ES /day yet it takes Much longer than 14 days to drop my target SNR by 3 db it has taken up to 2 months . still waiting for it to give me back another 3 15 days and counting ::)
Regards Jeff
Jeff, I'm fascinated.
Another possibility is that my reduction was earned by what DLM saw as 'stability' in my SNRM. That wouldn't be an unreasonable prerequisite for a target reduction. Now, my SNR wasn't particulary stable - fluctuating by 2- 3dB night to day. But thanks to my adverse tweak, it was always (for the past 7 days) an awful lot better than DLM may have expected it to be. I wonder if maybe that's more important than error rates?
7LM
-
Not having been around these forums for a while, :-[ I was fascinated to find this thread today. It's very frustrating that the workings of DLM are well documented in respect of sync, SNRM, IP Profile etc, and seem to more or less work, most of the time; yet the way that DLM handles the Target Margin seems shrouded in mystery, myth and magic.
I wonder if any of the techies here can make anything of my recent experience -
I'd been coasting along with my 52·5dB line syncing at around 5200, 6dB Target, until the cable just down the road was JCB'd by a careless contractor on 20 May. After the repair, I was down to 250kbps and a 12dB Target, and as expected the sync increased after 3 days, to 3776, with a low error count. As long as YouTube and BBC iPlayer worked, I was quite happy, and interested, to see how long it took for the Target to reduce.
Until then I'd always left the router on 24/7, so when nothing had happened by mid-August, I tried re-booting every day around mid-morning for a few weeks - still no change. At the beginning of October, I thought I'd better throw in the towel and ask Plusnet to request a reduction by BTw, but before I got round to it, I lost ISP connection around lunchtime on 6 Oct, due to a MUX failure in the local exchange. Sync was OK, but was at 576/288 with a margin of 31dB - just as though I was suddenly on a 0·5Mb fixed rate line! ???
The MUX was fixed by late afternoon, and I was back to 3776 at 12dB, but within a matter of minutes, the router re-synced of its own accord at around 4350 at 9dB. The following day around lunchtime, again without prompting, the router re-synced at 5120 at 6dB! :) :)
It would seem, then, that the DSLAM was stuck for nearly 6 months, and was somehow unstuck by a MUX failure.
-
Hi
Some people are just lucky I guess ;D keep fingers crossed
Regards Jeff
edit btw you don't happen to be on Anglesey?
-
Yeah - everything's crossed. ;D
Anglesey? No, Moray, NE Scotland.
-
Hi
quote until the cable just down the road was JCB'd by a careless contractor on 20 May they must have their mates down here :lol:
Regards Jeff
-
That's very interesting 4candles. One thing we have in common was that we had a period with a very high SNR (with low sync speed), that seemed to have preempted the target reduction. In your case however the reduction was very quick, so I'd say you're probably right about it being a result of the MUX failure.
One of the features I believe DLM implements is that if, as a result of a raised target, your line fails to sync, then the target will be reduced and your line will be 'flagged for investigation'. Maybe that's what happened to you, and maybe even resulted in human (common sense) intervention?
My own target has stayed contently at 9dB, despite the less than perfect stability that I expect with such a margin. It often drops to 6dB or so late evening, by which time error rates are usually 10-20 ES per hour. But in one 24 hour period earlier in the week I noticed around 100 ES per hour, and over 1400 ES in that 24 hour interval. That's the worst error rates I've ever seen but probably not unreasonable for an occasional occurrence, considering it's a 56dB line and synced at c3800, but I'm relieved to see DLM is no longer hyper-sensitive.
I conclude that, in my case at least, the SNR reduction that I got by request of my ISP was ineffective as it shot back up again after 48 hours. by allowing DLM to make up it's own mind to reduce, the reduction looks like it has been permanent.
edited: changed 'syncs below FTR' to 'fails to sync', which I believe to be more accurate.
-
One of the features I believe DLM implements is that if, as a result of a raised target, your line fails to sync, then the target will be reduced and your line will be 'flagged for investigation'. Maybe that's what happened to you, and maybe even resulted in human (common sense) intervention?
I conclude that, in my case at least, the SNR reduction that I got by request of my ISP was ineffective as it shot back up again after 48 hours. by allowing DLM to make up it's own mind to reduce, the reduction looks like it has been permanent.
edited: changed 'syncs below FTR' to 'fails to sync', which I believe to be more accurate.
I see what you mean, but sync was never lost during this episode, and as my initial 10 day training period completed before I understood such things, I'm saddled with an FTR of 2640, :( but don't get me started on that!
I think you're probably right re your own reduction - hope it stays down. :)