Print

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

[Ixel]

You're not the only one !

http://www.ewhurst-broadband.org.uk/?p=3528&cpage=1#comment-730

No doubt, but what has happened to me and others is not remotely similar to that report (other than experiencing a reduction).  There was no progressive reduction over time, just the immediate application of a lower banded profile in the face of a transient disturbance to the line.

IMO in my case it has nothing at all to do with FEXT or DPBO.  It is just a flaw in DLM which, for reasons we would all like to know, is simply refusing to back off on a perfectly respectable line with a BER currently estimated between 5*10^-11 and 2*10^-8, either of which are way above the alleged 10^-7 BER of a good line.

In other words, the profile is 'stuck' maxing out the profile max sync, while still only on an RCO of 73%.  It appears that both BlackSheep and Ryant704 were also stuck until their lines were reset, when their sync rose to the level their lines were capable of.  IMO, no one should be stuck on a profile which they max out, other than the current 80/20 exception.

While the product is sold as 80/20, I don't have any argument with a profile banding that results in a 79.997 actual sync on those lines that are capable of it.

Bit like mine, 60/20 instead of what I could get (unless interleaving kicked in) 80/20 due to the attainable being around 91/27. Tried getting BT Business to do what the engineer couldn't do due to lack of a special number that identified my circuit but it was more or less trying to pull teeth. Had a follow up call month or so later regarding my rating of the call being negative only to be told the same information and therefore wasting both my time and the person's time that phoned me back, I quote "Openreach won't reset it because your line is currently achieving a few megabits above your estimate, despite what the engineer may have showed or told you". I had wondered if I should've gone on the BT forum (business one that is) and contacted the moderators like some people do on the residential BT forum. But either way I imagine it'll require an engineer to come out, and given I have no fault I couldn't really expect one to come out just to reset DLM on my line without being charged a hefty fee.

[ColinS]

given I have no fault I couldn't really expect one to come out just to reset DLM on my line without being charged a hefty fee.

In all honesty I don't think they would do it even if you offered to pay them, as if it did 'fix' a stuck profile, it would open them up to the not-entirely-unjustified accusation that a 'stuck' profile was in fact the only thing wrong with the line.

It seems the perverse response to solving the 'up to xMb/s' complaints is to manage expectations downwards through the use of 'estimates'.  But why stop there, eh? Why leave your estimate at 60/20, when they could make it 15/2?  after all there's about as much justification given for either.  That would make life a lot easier for some people, but then it might just be open to a complaint that an 'up to 80/20' service was equally misleading.

And as you will find on other threads, there has indeed been a fair degree of 're-estimating' downward going on recently.

[Ixel]

given I have no fault I couldn't really expect one to come out just to reset DLM on my line without being charged a hefty fee.

In all honesty I don't think they would do it even if you offered to pay them, as if it did 'fix' a stuck profile, it would open them up to the not-entirely-unjustified accusation that a 'stuck' profile was in fact the only thing wrong with the line.

It seems the perverse response to solving the 'up to xMb/s' complaints is to manage expectations downwards through the use of 'estimates'.  But why stop there, eh? Why leave your estimate at 60/20, when they could make it 15/2?  after all there's about as much justification given for either.  That would make life a lot easier for some people, but then it might just be open to a complaint that an 'up to 80/20' service was equally misleading.

And as you will find on other threads, there has indeed been a fair degree of 're-estimating' downward going on recently.

True, well when my contract is up I'll probably move to Plusnet (cheaper for one reason), I assume with a migration the DLM will be reset again.

[Chrysalis]

The DLM can only sync to maximum rate of the band that has been applied (This is what the NGA Helpdesk told the engineer). My line had a profile assigned called TR03 (301) which was incorrect for my line, it was then applying a 7Mbps - 19Mbps band. It would keep syncing to the exact same sync (19Mbps with a 18.42 bRAS Profile) and band after full DLM resets, this went on for around 2 months. The NGA helpdesk then said it assigned a different profile (Shame they didn't tell me the name, was Kevin from BT care that did before and I was moved on to the ELC Team), the 7Mbps - 19Mbps band was no longer being used. This leading to my speed returning to the 25/27Mbps mark. Though 2 days later a 15Mbps cap was applied to my line from Wholesale... that's going off-topic though!

[EDIT] Chrysalis observed earlier in the thread

What are you speaking of is probably line banding which the new DLM does instead of adjusting target noise margins (I assume so it cant be overriden by people tweaking the noise margin CPE side).

To some extent I believe that this is so, but not entirely, as BE has also had fastpath removed.  But is that a temporary or a permanent change e.g. have BTOR updated the profiles recently?  I don't know but it is suspicious.

Simply using banded profiles with max sync rates like that is a very crude way of managing line conditions, but if they were doing that it might explain some things we are seeing e.g. reduced profile maxes while DLM still apparrently believes there is no need for either interleaving or INP.

The cynic in me wonders if BTOR are quietly banding everyone at the next nearest profile to the official BTOR 'Estimate'? 

Ironically its a way to claw out of BT infinity contract without penalty.

BT state "We wont slow you down"

Banding does exactly that. (interleaving does also).  i have already tested it, I was given a copout of my contract simply because DLM "slowed me down".  If they had to keep it legal, then everyone would have a max banded profile, interleaving would probably have to be able to be reset on user request also.

I've always agreed on this, I've asked BT this question. Here is the response (partly)...

"As you’ve mentioned, we can change the preference between Standard, Stable and Speed but in practice this will have little impact on the sync rate."

Then when the engineer was talking to the NGA helpdesk I got him to ask a couple of questions for me on the DLM (he was interested as well). He said there are many profiles that ISP use, none uses the default ones offered by Openreach, each ISP modifies the ones that Openreach offer though the default ones from Openreach will do exactly as they sound. He then mentioned you could see a speed difference of around 5Mbps if you're on the Speed profile instead of the Stable profile (note he wasn't talking about my line).

He then proceeded to talk on about Crosstalk, he stated if all people were on the "Speed" profile the service would degrade to a slower standard than if it was on a managed service. He said because everyone would be on "FastPath" there would be a higher number of errors this leading to more crosstalk? (First I've ever heard of this, I'm not sure if it's true as it wouldn't be the first time I've been lied to). Then proceeds to tell me that this would degrade the service to a slower rate if it wasn't managed the way it is.

When thinking about it a bit more, BT are sort of right.

Its actually openreach who slow down the service via DLM, the contract for the consumer is with BT retail, so BT can simply state we have not slowed you down in that regard.  So I guess I did get lucky when I was given a get out of jail card on my contract for DLM.

However an isp lieing to you about how vdsl works (eg. stating its acceptable for attainable speeds to drop off from 110+ to 65mbit) which in turn serves to lower expectations and accept the lower speed in a legal sense can be treated as slowing down the service and as such BT in breach of contract.  Just not so clear cut. 

errors themselves do not cause crosstalk, what causes crosstalk is the signal itself when it collides with another signal.  There are mechanisms to reduce crosstalk BT probably use such as twisting the pairs, and reducing dsl signal power, I expect they also vary the power pattern on adjacent lines to also minimise crosstalk.  Capping line speeds will only reduce crosstalk if the signal power is also reduced.  So eg. banding a line but leaving it on full power with a high snrm wont reduce crosstalk.  asbokid or someone else might correct me if I am wrong, but I think I am right.

[waltergmw]

@ Colin re our last two interchanges:-

When I last spoke to Azzaka probably around last Christmas he said quite categorically that the ONLY way of resetting a banded profile was to request a BT O site visit and that, provided the VDSL service passed all the line tests, that engineer was then able to request a reset, but NOT before. This particular conversation happened whilst I was supervising a Zen upgrade from 2 / 40 to 10 / 40.
Incidentally that Zen service is now sitting with another fixed Sync speed of 25.003 Mbps whilst another service two doors away from the same DP also on a single span drop is only running at 18.576 Mbps on a BT Retail service.
As you will expect that Zen user does not want to tempt the oracles by requesting yet another reset !

Kind regards,
Walter

[asbokid]

errors themselves do not cause crosstalk, what causes crosstalk is the signal itself when it collides with another signal.  There are mechanisms to reduce crosstalk BT probably use such as twisting the pairs, and reducing dsl signal power, I expect they also vary the power pattern on adjacent lines to also minimise crosstalk.  Capping line speeds will only reduce crosstalk if the signal power is also reduced.  So eg. banding a line but leaving it on full power with a high snrm wont reduce crosstalk.  asbokid or someone else might correct me if I am wrong, but I think I am right.

That does sound right, Chrysalis.  I'm sure you've studied this for longer and know much more about the subject than me though!   It is a good question:  Does limiting the maximum transmission rates through the DSLAM channel-profile, in itself limit crosstalk?  Is data (if only sync data) still modulated on those unused tones and unused decision points in the QAM constellation maps?  If there is no modulated data on a tone, does an unmodulated carrier signal by itself contribute to crosstalk?   

cheers, a

[Chrysalis]

Its a good question just I think 'any' signal whether used for data or not is noise to another line.

If we look at the hg612 graphs eg. there is a graph that shows the snr, not snrm but the entire snr.  that is the DSL signal.  If a line is banded that snr does not decrease, it stays the same size. A line downgraded to 40/10 will have the same snr as it did on 80/20. As BT keep it on profile 17 and I believe do not reduce the power.  that snr is available for use by the modem/dslam at anytime eg. bitswapping can move bits around to utilise spare snr.

This may be wrong its just how I see it.

[ColinS]

Well, I hate to disagree, but I don't think that's entirely right.  Short of Vectoring I don't think the DLM will vary power patterns on adjacent lines other than DPBO & PSD masks.  I do agree that banding the line is unlikely to reduce crosstalk much, if at all.

However, I have come across some recent research that might form a basis for determining if crosstalk levels are or are not significantly affecting a line.  Can't remember the guy's name at the mo, but I think it began with an A. 

Anway, his research is here: http://forum.kitz.co.uk/index.php/topic,12300.45.html @ post 48

[ColinS]

When I last spoke to Azzaka probably around last Christmas he said quite categorically that the ONLY way of resetting a banded profile was to request a BT O site visit and that, provided the VDSL service passed all the line tests, that engineer was then able to request a reset, but NOT before. This particular conversation happened whilst I was supervising a Zen upgrade from 2 / 40 to 10 / 40.
Incidentally that Zen service is now sitting with another fixed Sync speed of 25.003 Mbps whilst another service two doors away from the same DP also on a single span drop is only running at 18.576 Mbps on a BT Retail service.
As you will expect that Zen user does not want to tempt the oracles by requesting yet another reset !

Hi Walter.  Yes, I understand the points you are making. 

However, it would help to know whether either of these two lines are maxing out their current profiles and/or capability or not?  i.e. what are their respective RPO & RCO %s?  Do you know?

[waltergmw]

Hi Colin,

Sadly no, as both are ECI modems on to an ECI DSLAM.
I've not yet been brave enough to try a different extraction method.
I might be able to persuade the faster one to let me disturb his connection with an unlocked Huawei, but he won't thank me for screwing up his quite reasonable connection for this line length.

Kind regards,
Walter

[ColinS]

Sadly no, as both are ECI modems on to an ECI DSLAM.
I've not yet been brave enough to try a different extraction method.
I might be able to persuade the faster one to let me disturb his connection with an unlocked Huawei, but he won't thank me for screwing up his quite reasonable connection for this line length.

Sure, no problems.  I was just curious.

[snadge]

just read through most of this...

been a good read, I think I understood how it works as soon as BS said it (as it seemed the only logical way) and as I read on i was confirmed right   , the "banded profile" must be determined by DLM at time of connection, from the line stats it determines the best BP for that line and assigns it, connecting anywhere from the home to DSLAM the sync output is the same, totally understand that - but if you were to shift the house back (quite far) then DLM would see that it cannot operate a connection on that BP anymore and must lower it....and anywhere from the 'shifted house' to the DSLAM you would be the same new lower profile..  ...least thats how I see it.

what concerns me is the amount of crosstalk that 'appears' to be picking up as more and more VDSL connections go live - it kinda makes me hope we DONT get FTTC if all ive got too look forward to is watching my 40Mb connection slowly dwindle away to 'slightly better' than top ADSL2+ speeds...  whis is why they need to get Vectored VDSL installed ASAP... i heard Alcatel-Lucent had a design ready that didnt require any extra hardware installed (I think) just fw updates to customers routers (or was it customers needed new routers but PCP equipment didnt change?? cant remember)

[Bald_Eagle1]

what concerns me is the amount of crosstalk that 'appears' to be picking up as more and more VDSL connections go live - it kinda makes me hope we DONT get FTTC if all ive got too look forward to is watching my 40Mb connection slowly dwindle away to 'slightly better' than top ADSL2+ speeds... 

My DS connection speed has reduced from around a sustainable 30Mbps (after my line was finally repaired) to less than 21Mbps - strongly suspected to be as a result of increased crosstalk.

Although that is very disappointing, due to distance (from the exchange for ADSL connections), it is still far better than the 1Mbps that my connection managed on ADSL.

[les-70]

  If there is no modulated data on a tone, does an unmodulated carrier signal by itself contribute to crosstalk? 

   I can't say for VSDL  but my ADSL line has some odd behaviour w.r.t one my neighbours which suggests that for ADSL it make no difference whether there is data on the line.

    I have been able to phone up and ask 6 neighbours if they would switch their modems on and off for me.  Five give a change in snr of less about 0.1db each but one neighbour gives  a drop of 1.4db.  That drop and the matching change in QLN do not vary at all once they are synced.   Apart from this drop of 1.4 the biggest single incremental change seen on my line is about 0.3db.   That 0.3db drop does not seem to be an immediate neighbour.    The peak to peak change ever seen in snr is a bit less than about 3db with half from one property!

  I don't know how bad an effect a spit pair can cause but I  wonder whether I share half of a pair with the offending neighbor.  Might this possible??

[asbokid]

When I last spoke to Azzaka probably around last Christmas he said quite categorically that the ONLY way of resetting a banded profile was to request a BT O site visit and that, provided the VDSL service passed all the line tests, that engineer was then able to request a reset, but NOT before.

Hi Walter,

That must be down to BT's rules-and-regs, rather than any technical limitation? The DLM software runs centrally at some NOC (Martlesham?).

Every DSLAM has an inband management interface. The DSLAM, and each of its subscriber ports, are configured and re-profiled through SNMP by the (Java-based?) DLM application. 

So a DLM reset is all software-controlled, and requires no "truck roll".

Wonder why BT opted primarily for site visits, and the great cost they entail?

cheers, a