Print

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

[HPsauce]

OK, here's the topic for discussion, and I genuinely have no idea what the answer is.

Lots of people seem to have issues where BT's kit reacts very quickly to a transient, or doesn't react to fixing a long-term problem and they have an inappropriately high target SNRM. Sometimes this appears to be stuck and never moves. Sometimes it moves but very very slowly.

I guess this problem MAY occur on LLU kit, but it seems to be common on BT.

Many try to overcome this by forcing it lower, e.g. with DMT or modified router firmware. A maximum move of 6dB appears to be possible at present.

The question is, does this forcing actually have long-term benefit?

Obviously in the short-term it can/will give better speeds, but what does BT's kit see and how does it react?
I can, for example, envisage that a "forced" line will have a higher error rate than an "unforced" one so, depending what the algorithm is, forcing may slow down the automatic adjustment process.

Thoughts anyone, or even better real insider information....

[roseway]

From my own recent experience I don't think that forcing the target noise margin down is, of itself, likely to obstruct the automatic procedures. But fairly obviously, if it results in a higher error rate, then this can have an effect. I really have no hard information, but I suspect that the best approach is not to force down the margin, and have patience, but it's hard advice to follow.

[sevenlayermuddle]

I'm convinced it hinders.  But I stress I'm by no means I'm an insider, just trying to make educated guesses like everybody else...

According to a BT patent application for an ADSL DLM (which may or may not be the current DLM),  each line is categorized, based on errors and frequency of resynchs, as being 'very poor, poor, acceptable and very stable'.  That patent goes on to describe how, in order to stay level in terms of target marging, the line needs to be 'acceptable'.  But in order to win a reduction, the line has to be 'very stable' - i.e. better than it was to begin with.

Now, if you sit back and suffer the consequences of a high target margin, your line probably connect at a slower speed so you ought to get fewer errors, and may well be judged to be 'very stable' in which case you'd get a reduction.  But if you tweak it back again to the old connectioon speed, it will have the same error rates, which have already been decreed to be merely 'acceptable' and so the margin won't move.

I stress that Im guessing.   I've no way of knowing for sure whether that BT Patent describes the current DLM, or some other DLM.  But it makes sense to me.

[broadstairs]

I am not sure how many LLU ISPs use any kind of dynamic management. I dont believe TalkTalk does because of the number of people who have to ask for their profile to be changed, I know they can look at your stats but I think the only time they do is when you ask for a change.

[sevenlayermuddle]

I am not sure how many LLU ISPs use any kind of dynamic management. I dont believe TalkTalk does because of the number of people who have to ask for their profile to be changed, I know they can look at your stats but I think the only time they do is when you ask for a change.

That's a pity, as BT's DLM serves a useful purpose. The majority of 'Joe Public's migt suffer in silence for an awful long time before complaining if the line's flaky, and BT's DLM should take care of things for them.

The one big failing (IMHO) of BT's DLM is that it can't tell the difference between reschys that are cause by flakiness, and resychs that are caused by the user resetting his router.  As so many of us know, reset the darned thing too many times and up goes your margin.  Even if you explain things to your ISP, BTw only have your word for it that's what caused the problem and so may refuse to override it.

[waltergmw]

Perhaps the saddest thing is the culture we observe within BT O which makes it so difficult for us and our ISPs to achieve changes efficiently. The up-side is that this site thrives as a priceless haven of research !

At a lower level, and certainly excluding current company, I find it quite a shame that so many ADSL users are oblivious of ADSL techniques and accept whatever speed they find is provided. Many are not even able to say whether they are on a fixed or rate-adaptice service. Consequently many suffer in silence not even recognising the benefits of both higher performance and personal contentment.

Kind regards,
Walter

[kitz]

Some very good comments and I dont disagree with a single point made so far.

Re over-riding it yourself, although Ive no hard evidence, I would suspect that if you push it down too far and therefore rack up more errors then it may take a bit longer.  Its swings and roundabouts..  what for what may take ages... or tweak and delay the natural process a while, but in the meantime at least you are getting the speeds your line is capable of.  If I were put in the position on my line, then I'd still tweak.

Why BTw seem to make it hard for ISPs to request profile changes Ive no idea, I know other wholesale LLU providers who make this a fairly easy process by giving the ISPs a control type panel so they can change the customer DLM profile.  Then you have some other LLU providers such as Be who even let the customer do it themselves.

[sevenlayermuddle]

Kitz,

The point I was making was that if, by tweaking, your error rates remain merely 'acceptable' then the margin may never, ever change, no matter how long you wait and indeed some people (me!) seem to have experienced that phenomenon.

My hypothesis (based on that patent) is that the margin will only be reduced (and then, only after a long delay) if the error rates make the transition to 'very stable', which is only likely to happen if you are willing to endure a slower connection speed.  The patent paper doesn't specify how long the 'very stable' rate needs to be sustained for in order for the target to drop, though it mentions "a week or a month".

I'm tempted to put it to the test by CLI tweaking it up to 21dB, and then leave it there, driving in the slow lane for at month to see what happens.  Maybe I'll try it next time I go on holiday, and leave the router switched on.

[kitz]

>> The point I was making was that if, by tweaking, your error rates remain merely 'acceptable' then the margin may never, ever change, no matter how long you wait and indeed some people (me!) seem to have experienced that phenomenon.

Yes I accept that is  a very valid point and something I had also wondered about.  The lack of information though makes it hard to know just what really does go on :/

>> I'm tempted to put it to the test by CLI tweaking it up to 21dB, ../snip/...  Maybe I'll try it next time I go on holiday, and leave the router switched on.

lol - thats an idea