Kitz Forum
Broadband Related => Broadband Technology => Topic started by: jid on September 07, 2013, 09:28:11 PM
-
Admin note:- This thread has been split from jid's original thread (http://forum.kitz.co.uk/index.php/topic,12892.0.html) so that the issues of BToR's DLM can continue without distraction from the main topic.
This may give the answer: http://www.thinkbroadband.com/news/5929-reminder-about-the-reality-of-dlm-on-openreach-fttc.html (http://www.thinkbroadband.com/news/5929-reminder-about-the-reality-of-dlm-on-openreach-fttc.html)
The Openreach Fibre to the Cabinet products use a DLM system that has three main modes, the only control a retail provider has is to pick which of these three profiles your new FTTC connection will run on, or swap you between the various modes.
Standard – best overall balance between speed and stability for general internet users.
Stable – prioritise stability over speed for IPTV videoconferencing, home workers and businesses transferring data and IPTV.
Speed – prioritise speed over stability for online gamers.
-
This may give the answer
To what Jamie? It's not actually known whether or not these three so-called 'profiles' are made available to ISPs, but if they were, then what you assume may be right - but it wouldn't explain why your bit loadings change as much as they appear to do.
-
It's not actually known whether or not these three so-called 'profiles' are made available to ISPs, but if they were, then what you assume may be right
They definitely are. These are the only controls the ISP has over the BToR DLM.
ISP's usually have a default setting for when the line is provisioned on fttc, but this can be changed at a later date if they wish at EUs request.
Take Zen fo eg who by default provision on the speed profile, they will happily change your profile to Standard or Stable.
-
They definitely are. These are the only controls the ISP has over the BToR DLM.
Then I stand corrected Kitz. :-[
Were you asked by PN then which profile you would prefer when you signed up for FTTC, because I wasn't? If not, then on what basis do they select among them, and how are these mapped down onto the 192 line profiles available to a Huawei DSLAM? That would be interesting to learn. :-\
[Edit] Thinking back, I far as I had been previously aware the 'source' reference to these 3 'profiles' was originally in the DLM patents filed by BT, which, as others have observed, originally applied to ADSL. However, there are still strong similarities to those patents in today's VDSL DLM. In addition I seem to recall that these 3 profiles are still echoed somewhere in the GEA SIN (see the reference provided by Ryan below), but I hadn't previously been aware that these were being actively selected by ISPs on BTOR FTTC. Perhaps all I need to do then to get rid of my 1233 DS Interleaving depth is just to ask PN for them to put me on the 'Speed' profile (i.e.fastpath)?
If I recall those patents correctly, the 3 different profiles were used to describe a) different (potential) levels of errors and or forced resync rates when DLM was monitoring the line and b) subsequently more or less aggresive (than standard) action on the part of DLM when deciding whether to take or relent from action on the line.
-
>> Were you asked by PN then which profile you would prefer when you signed up
Nope.. I suspect that like Zen they also opt for the default speed profile. They should be able to get it changed easily enough if you want one of the others for say stability. The option is there. Its up to the ISP to choose what default they want. Azzaka could probably provide more info. Not all ISPs will admit its actually there ;)
Yes you are correct that they do practically follow the same guidelines talked about years ago for adsl.
In a weird way its a more simplistic version of the BE DLM.. but then they make it more complicated & interweave added parameters on top of this base profile so that things like interleaving will still kick in - even on the speed profile.
I can perhaps understand why the ISPs dont make it obvious to the EU. If the option is there, most people wouldnt have a clue about such things & would likely select speed anyhow.
---
Edited to add because I forgot to answer:-
Perhaps all I need to do then to get rid of my 1233 DS Interleaving depth is just to ask PN for them to put me on the 'Speed' profile (i.e.fastpath)?
They should certainly be able to check which profile you are on. Moving to speed does not turn off interleaving. That cant be turned off permanently and the ISP has no control over it once the BToR DLM decides you need it. :(
-
Nope.. I suspect that like Zen they also opt for the default speed profile. They should be able to get it changed easily enough if you want one of the others for say stability. The option is there. Its up to the ISP to choose what default they want.
Well, for the moment, I hope you'll forgive me if just for now I remain unconvinced about this - not that I don't believe what you are telling me. I would genuinely like to see the stats of anyone who has been on one of these profiles on VDSL and what it changed to when or if they ever got moved to another. Is there anyone out there with that information or experience of doing that?
[Edit] I know which profile I'm on, but it is one of the 192, not one of 3!
and with that last thought he drifts off to :sleep: :)
-
just a reminder in case you didnt notice... I added something to the above post about your request.
I hope you'll forgive me if just for now I remain unconvinced about this - not that I don't believe what you are telling me.
nps.. ill find a link for you.... (sorry took a bit longer than I thought to find it)
https://forum.zen.co.uk/forums/permalink/43293/43339/ShowThread.aspx#43339
-
The DLM profiles are Stable, Speed and Standard. [1]
The 192 profiles are a line profile which are separate from the DLM profile I believe? I know BT set the DLM to "Standard" by default (at least that's what I was told).
[1] http://www.openreach.co.uk/orpg/home/products/super-fastfibreaccess/fibretothecabinet/fttc/downloads/GEA_FTTC_4.pdf
-
Thanks to all for widening the debate on this. I am almost certainly completely wrong about this, but, personally, I haven't seen any actual evidence that provides a convincing narrative (of how these 3 profiles are used) on VDSL, other than in how DLM might monitor and subsequently react to line conditions.
Yes, I have seen this repeated many times on the net (and always in the same way), but I have yet to see anything that confirms that it actually works 'that way'. I have seen exactly what Ryan quotes, many times, but never with any evidence of how it subsequently influences VDSL DLM. The Zen document, IMO, is merely repeating what the BT GEA SIN itself says, and AFAICS, that is still echoing the original statements in the BT DLM patents (for ADSL). Obviously though, I can see analogies with the 3 'profiles' that were available to EUs as former BE ADSL users.
So, I have no problems with understanding in principle that an approach like that could have been adopted for VDSL too. It's just that so far I can see no evidence of how that works in practice.
I can also understand what you are suggesting Kitz when you say
In a weird way its a more simplistic version of the BE DLM.. but then they make it more complicated & interweave added parameters on top of this base profile so that things like interleaving will still kick in - even on the speed profile
I suppose it's that interaction that I do not, as yet, follow.
I also agree with you, but for a different reason I suspect, when you say
interleaving ... cant be turned off permanently and the ISP has no control over it once the BToR DLM decides you need it
So, being >:D advocate here, what exactly do we think that such a 'speed' profile might cause DLM to do, if it's not to avoid the additional delay of using the RS+Interleaving error recovery technique? After all, in all the places that refer to it, it is said (even by Zen) that it is to
prioritise speed over stability for online gamers
And conversely, imagine for the moment I'm on a Speed profile (and a perfect line), and I request that my ISP moves me to a Stability profile instead. Would that switch Interleaving on? I would suggest that it probably would. So, ISP's could then, in effect, have Interleaving switched on (if indirectly), but then lose control of it, because they then cannot switch it off again by restoring the EU to the original Speed profile on which (s)he was on just moments before?
Perhaps I am missing something fundamental here, but that all seems irrational to me. But it wouldn't be the first time that I had missed something obvious ....
As a counter-example, we all know from what BS has confirmed, that everyone is initially put on a so-called 'open profile' (e.g. of 80/20, fastpath) on activation, and that (unless things are really bad) DLM does not intervene to change that until at least 24hrs later. How does the fact that we are all activated in this same state sit with the ISP selection of one these 3 profiles, where according to common belief, Zen selects the Speed profile, while BTR selects the Standard profile?
Finally, PN have told me previously that my profile is '40M-80M Downstream, Interleaving Low - 10M-20M Upstream, Interleaving Off', which, as I have suggested, is one of the 192 line profiles used by the Huawei DSLAM. No mention anywhere of Stable/Standard/Speed, nor any inkling of how these 'DLM profiles' are translated by DLM to the applicable line profile.
Anyway, that's just the basis of my (hopefully healthy) scepticism about this. ??? As I have already said, I just don't understand how it is supposed to work as yet. :no:
What would be really interesting would be if anyone can actually request a change between any two of these 'profiles', and the differences can be recorded. We have the tools available to see that.
-
@ Colin,
Just to add a little verisimilitude to
initially put on a so-called 'open profile' (e.g. of 80/20, fastpath) on activation
even on a quite atrocious line.
I observed a VDSL installation by a subcontractor in the late afternoon of day 1 with a downstream sync of 39.99 Mbps and an upstream IP profile of 2 Mbps.
That remained throughout day 2 up until at least 21:20 but before 08:09 on day 3 it had settled down to a nice steady sync 4.38 Mbps down and still an IP profile of 2.00 Mbps up.
YES that IS 4 decimal 38 down !
It has remained around that performance ever since.
Unsurprisingly it is still on my list for some attention in due course.
Kind regards,
Walter
-
Thank you for that Walter. :)
Forgive me though if I'm not completely following the important point I'm sure you are making. :-\ I can be a bit slow on the uptake sometimes. :(
In your example, I assume that the product was 40/2? So in that case an open profile would be 40/2, fastpath. Which is what it seems to have achieved on activation in the late afternoon of day 1? If DLM doesn't interfere for (at least) the first 24hrs, experience tells us that thereafter it generally intervenes (by changing profile) in the early hours of the morning (say ~5am +/-). Which would seem to fit your timescale of between 21:20 day 2 - 08:09 day 3, by which time it had sadly taken exception, it seems, to whatever it observed during that short period. :(
Have I got that right, and am I right in guessing that your point is that even 'bad' lines (that perhaps might have merited activation in a 'stable' profile) are still activated in an open one? :-\
(Sorry though if I've misunderstood :))
but Good luck with the fix anyway!
-
Colin ok the 3 DLM profiles are seperate to the DSLAM profiles.
DLM is not run on a DSLAM, it just controls DSLAMS.
As you said regardless of DLM profile everyone starts off the same way, the reason is these profiles dont affect how you start, they just affect how much instability is required for you to move to a lower profile.
So eg. on speed it might require 1000 CRC errors a minute, standard might be 600 a minute and stable might be 300 a minute.
-
Wanted to comment on this yesterday - not had chance so very rushed post coming up, including part of a PM
This is also the sort of thing that if we do manage to get some information together is put something either on the main site or wiki where it can more easily be indexed. ie What the DLM says it should do.. and what the general observations are in reality.
I linked to the Zen thread, because Zen are one of the ISPs who are happy to submit a change of profile.. Ive actually seen them say it elsewhere too, and they will check and state which BToR DLM you are supposed to be on rather than give the 192 line profiles.
It would be interesting to know if a line goes through the speed -> standard --> stable process as part of the 192 line conditions.. and if once you are on standard then it wont automatically go back down to the 'speed' range.
Kind of like how in the early days of the Maxdsl DLM. All lines would start as interleaved off, but once it was triggered then it was hard to get it turned back off even when the ISP would set the line to off.
What Ryan/Chrysalis says put some of the pieces in place... and perhaps much of the confusion has come about because we tend to use the term 'DLM' when there is obviously the base DLM profile and also a Line profile.
In the same way that I mentioned in one of my earlier posts that its almost like Be's DLM, but then theyve woven more parameters on top.
Everyone starts off on open to give a line chance to see if it needs intervention or not or even how much ... and if theres a problem then DLM kicks in
What those parameters are depends upon the basic DLM profile what the ISP can select... and then we end up with the line profile.
What isnt clear is what effect an ISP selecting a lower DLM profile would actually have on the resultant line profile. It may not remove interleaving but would it say reduce the depth.
Can anyone provide the link to the 192 profiles please.. I cant seem to find it atm.
It would be interesting if BToR have just selected some of them. Because if its the same profiles that Tiscali used to use on their Huawei MSANs then I believe they may have done... because from what Ive seen they dont seem to use SNRm for stability - instead opting for interleaving, leaving SNRm at 6dB.
If Zen was selecting speed then I would at least expect some lines to show a 3dB SNRm? So it would appear that they have disregarded any SNRm options, which is a pity as Im sure there are some gamers who would happily sacrifice some speed in exchange for better latency if their line was a bit wobbly.
What would be interesting is if someone could go through the DLM change by the ISP and then see what effect it had on the line profile.
------
PS Has anyone seen any target SNRms on FTTC other than 6dB?
If so can they recall where.
-
Can anyone provide the link to the 192 profiles please.. I cant seem to find it atm.
It would be interesting if BToR have just selected some of them.
As far as I can recall, it was Asbokid who mentioned that a Huawei MA5616 MSAN has the potential for usage of up to 192 profiles and that Openreach only use a small selection of them. :-\ We shall just have to patiently wait for A's return from his summer sojourn.
-
I concur with B*Cat. :) A discussion started here http://forum.kitz.co.uk/index.php/topic,12535.0.html (http://forum.kitz.co.uk/index.php/topic,12535.0.html), which includes a short list provided by PN that B*Cat republished. Then at some point, as a consequence, Asbo did (I'm sure he did :-\) send me a list in a private communication. But can I find it now? :no: Anyway, I'll keep looking, but when Bald_Eagle returns, I might ask him, as I seem to remember sending him a copy at some point (or is my memory just running away with me :-X).
It's always handy to know someone with their own DSLAM to ask! ;D
-
Colin ok the 3 DLM profiles are seperate to the DSLAM profiles.
DLM is not run on a DSLAM, it just controls DSLAMS.
As you said regardless of DLM profile everyone starts off the same way, the reason is these profiles don't affect how you start, they just affect how much instability is required for you to move to a lower profile.
So eg. on speed it might require 1000 CRC errors a minute, standard might be 600 a minute and stable might be 300 a minute.
Yes, thank you. This is paraphrasing how it was described in BT's original DLM patents for ADSL - EP 2 169 980 A1 & EP 2 342 902 B1 (original source 7LM) - and was what I was citing when I referred to how DLM might monitor and subsequently react to line conditions.
These are referred to as Stability Policies, as discussed at file 1 below, and is, as I said, IMO the source of these 3 profiles.
The actual metrics suggested and described therein are at file 2 below, and an overview of the original DLM algorithm at file 3 below.
What appears to be different in the VDSL algorithm is that these have to be mapped down to the selection amongst & application of (one of) the available set of line profiles in the DSLAM, each of which in themselves embodies (to use the terminology of the original patent) a set of parameters whose selection, as Kitz has referred to, adopt a distinctly different (so-called banded) approach to that contemplated (and used) by the original ADSL DLM(s).
What you have partially described are what might trigger a profile change, not what profile (and so parameters) are selected, or why, or how these parameters are related to a putative set of stability policies.
It is likely that there are multiple profiles with the same banding, but with different parameters, e.g. INP and delay, which could be selected between on the basis of these stability policies. However, it would be nice to see some evidence that e.g. someone on a so-called Speed (stability) profile encountering line issues is simply moved to a lower speed banded profile, but one that does not use INP (because of the delay introduced).
It could be argued that when issues on my line were caused by excessive forced resyncs, it behaved in that way by simply applying a lower banded profile i.e. as if I was on a Speed profile; but when subsequently issues arose which were the result of excessive error-rates instead, it behaved otherwise, in selecting a profile in the same band, but with INP and delay instead.
So, to be >:D advocate again, what is differentiating the approach here - the nature of the trigger, or the 'applied' stability policy (which must have been selected, or defaulted, by the ISP, and remained unchanged)?
-
Kitz there is no 3db snrm profile used by BT on FTTC. None of the profiles as far as I am aware directly control SNRM at all. Instead they use banding to cap the speed, this of course increases the SNRM but its done differently to older DLM which increases the target SNRM instead. BT do seem to prefer to apply interleaving prior to using banding tho, I dont think I have seen a single instance of a line that has been banded whilst still on fast path, its possible a line can recover from interleaving to fast path and still be banded (my line was) but I have never seen BT apply banding prior to enabling interleaving.
I have seen zen (and I think also plusnet) reps state the 3 DLM profiles isp's can choose from have all the same dslam profiles available to them, the only difference between the 3 DLM profiles is when a line is considered stable or not, they have different thresholds, thats it. So its entirely a possible a line can be on the stable profile but with a 80/20 sync on fast path and a line can be on the speed profile banded down to 20/10 with a high interleaving depth.
Colin your images seem to confirm what I just said, but where are they from?
-
I dont think I have seen a single instance of a line that has been banded whilst still on fast path, its possible a line can recover from interleaving to fast path and still be banded (my line was) but I have never seen BT apply banding prior to enabling interleaving.
Then you need to pay a little closer attention to what I said above. Not long after I moved to FTTC, work was done on my D-side (but not necessarily on my line). As a consequence I then spent the next ~56 days being moved down and then back up a series of banded profiles before finally returning to 80/20, and during the whole time I remained on fastpath. I have the evidence for that (thanks to BE), where is yours?
I have seen zen (and I think also plusnet) reps state the 3 DLM profiles isp's can choose from have all the same dslam profiles available to them, the only difference between the 3 DLM profiles is when a line is considered stable or not, they have different thresholds, thats it.
I can't speak about Zen, as I don't know, but PN are still learning about FTTC and know no more about it than anyone else. One of their staff volunteered to try to find out more about it in response to a question I posed over there about 'how it works'; he said he 'thought that was his job' (to find out). Sadly, he couldn't or didn't, and he certainly didn't mention at any time anything about these 3 stability policies. If they are indeed still in use, and as you suggest the only difference between them are the trigger thresholds, perhaps you could explain to me the 'benefit' to a gamer of being on a 'Speed' profile if INP and Interleaving get set resulting, e.g. as on Jamie's line, in an additional delay of 13ms?
So its entirely a possible a line can be on the stable profile but with a 80/20 sync on fast path and a line can be on the speed profile banded down to 20/10 with a high interleaving depth.
and it's also entirely possible that it doesn't quite work the way it used to with ADSL.
Colin your images seem to confirm what I just said, but where are they from?
They are from the original ADSL DLM patents filed by BT, and identified by 7LM quite some time ago I believe, on this forum.
I started this discussion by acknowledging that this is what people believe (an urban myth), but explained that I remain as yet unconvinced. If something gets repeated often enough, people tend to start to believe it without question. There are plenty examples of that. Unfortunately, you have not provided any real evidence that supports this theory. Most of what you have referred to, and 'confirms' what you said, actually reflects BT's original patents for ADSL DLM.
Whereas I can provide actual data from my own experience (thanks again to BE) that, to the contrary, it appears that there is more evidence that VDSL DLM has possibly 2 different 'stability policies' (which may be built-in) - one where the dominant line characteristic observed by DLM is instability (excess forced resyncs), and the other where it is performance (excess error-rates). So, for the moment, I'm going to continue to suggest that 'your Emperor has no clothes on'. :)
-
I have to disagree the DLM doesn't work with SNRM values in mind, my experience has been the DLM will always increase/decrease the speed depending on the current SNR Margins. Though this is more indirectly but it's the only way it can control the SNRM (to a degree).
I'm currently banded at 20Mbps on Fastpath due to a blue transformer exploding the other week. Before this my line would sync around 25Mbps with a INP of 3 and a Interleaving depth of 450. When I was first banded it was on Interleaving but I was still hitting my normal 60/70 (far better than the start of the service, use to have up to 2000/3000) FEC per second as I suffer from REIN. Been on FastPath now with FEC enabled and hitting about 10/20 with the same in CRC.
I think it's key we try and leave the ADSL DLM Information out of this discussion, any Information posted should be about the VDSL2 OR DLM as they're entirely separate from BT Wholesales. I will contact BT and ask if the DLM for BT swaps around the 3 profiles and ask for my current profile status as I'm assuming it will be on 'stable'.
Also the majority of you seem to be on Huawei cabs but I'm on ECI, I'm assuming this makes no difference to the DLM but I wouldn't be surprised if it has little changes for each vendor.
-
Thank you Ryan, that's an important contribution to the discussion, and one which fairly convincingly provides further evidence that, however it works, it no longer works quite the same way it used to do.
I would agree with you, that you would hope, assuming a centralised DLM, that its behaviour would be fully independent of the FTTC cabinet DSLAM-supplier, but I think (or at least seem to recall), as you suggested it might, that some users on ECI cabinets have reported slightly different behaviour, but I am happy to be told otherwise. It would be very helpful if any such users could report their own experiences here too like you.
-
I concur with B*Cat. :) A discussion started...
Thank you b*cat & Colin for trying. :)
Since then I've found something that would indicate that whilst Huawei DSLAMs have line profile templates, the SP can very easily delete/over-write, amend or adjust any of these profiles. The doc I found was partly obscurred but it clearly indicated how it was easy to change these profiles so its likely a moot point as I bet BT have been tweaking to their own anyhow.
I have a page somewhere showing the old Tiscali profiles that they used on Huawei MSANs
What appears to be different in the VDSL algorithm is that these have to be mapped down to the selection amongst & application of (one of) the available set of line profiles in the DSLAM, each of which in themselves embodies (to use the terminology of the original patent) a set of parameters whose selection, as Kitz has referred to, adopt a distinctly different (so-called banded) approach to that contemplated (and used) by the original ADSL DLM(s).
Agreed. This is what I was trying to suggest when I compared it to the BE options only with more complicated interleaved parameters weaved on top based on the original selection.
Thanks to the patent links it would seem that this is correct. Its the ISP (rather than the EU) which is choosing what I shall call the base DLM profile. Then depending upon the base DLM profile, differing parameters set the line profile. In other words some of the base profiles are more tollerant than others and require more Err Seconds or re-trains per hour to trigger a higher line profile.
It is likely that there are multiple profiles with the same banding, but with different parameters, e.g. INP and delay, which could be selected between on the basis of these stability policies.
There could be and certainly this would be the logical assumption.
However - Ive not read the patent in minute detail... but it points to more towards tollerance for the parameters... and what is classed as 'unstable' by the DLM system.
It appears to work almost backwards to how youd expect a straight forward DLM to work - more on this later!
If this is the case, (& because the ISP cant reset) then switching to a different base profile could in theory reduce a line profile faster than others.
-
Kitz there is no 3db snrm profile used by BT on FTTC. None of the profiles as far as I am aware directly control SNRM at all.
Thats what I said, all the evidence seems to be a 6dB target SNRm regardless and I'd not seen any evidence to the contrary
"they dont seem to use SNRm for stability - instead opting for interleaving, leaving SNRm at 6dB"
If they had been then I would expect at least some Zen users on speed profile to see a 3dB target SNRm, but we haven't. We may have seen it on BTw's 21CN, but there's a complete lack of this profile on BToR's line management. We do see however seem to see more line banding on fttc.
I have to disagree the DLM doesn't work with SNRM values in mind, my experience has been the DLM will always increase/decrease the speed depending on the current SNR Margins. Though this is more indirectly but it's the only way it can control the SNRM (to a degree).
Thank you ryan, the BToR DLM doesnt seem to set target SNRm, but like you say in effect controls this through banding.
iirc the BTw system only used banding when it had gone through all the SNRm and interleave options.
The BToR DLM appears to make more use of banding rather than target SNRm. By banding a line it is in effect increasing the available SNRm.
The old Tiscali DLM used to use banding too, but as well as a DLM change it could also be easily manually configured by the ISP.
I will contact and ask if the DLM for BT swaps around the 3 profiles and ask for my current profile status as I'm assuming it will be on 'stable'.
Thank you.
Also the majority of you seem to be on Huawei cabs but I'm on ECI, I'm assuming this makes no difference to the DLM but I wouldn't be surprised if it has little changes for each vendor.
No difference Im also on an ECI.. Its why I wasnt too concerned about not getting info on the default Huawei profiles. Its extremely likely BT have over-written any Huawei and ECI defaults and set their own that will match for both cabs.
-
Colin whats this asking for proof business? isnt a competition. I said I didnt recall seeing an instance of someone been banded without interleaving, so you are one such case, but you are still the only case I know off, so it doesnt really change anything that the DLM system seems to heavily lean towards interleaving as a first line mechanism for line stability. I have read dozens of forums, 1000s of posts from many many people discussing DLM, FTTC etc. and out of all these publicised reportings you are the single one who had a line banded prior to interleaving, so yes it can happen as you have proven, but I wouldnt say it trashes the idea that DLM heavily favours interleaving.
Also regarding the DLM profiles this is in regards to FTTC not ADSL, I dont read up ADSL things and assume its FTTC I have little interest in ADSL. The 3 profiles is very much so in reference to FTTC.
perhaps you could explain to me the 'benefit' to a gamer of being on a 'Speed' profile if INP and Interleaving get set resulting, e.g. as on Jamie's line, in an additional delay of 13ms?
Thats a question to BT not myself, gamers dont even benefit from speed, they thrive on low latency and jitter. In fact I believe openreach just call it a speed profile whilst its the isp's who have decided to label it for gamers. I didnt design the DLM system I am just commenting on how I believe it to work based on openreach documentation, customer feedback and public comments made by isp's.
If people on this site are going to continue to just disregard what I am providing then I will simply stop taking part, no skin of my back.
Just because your line behaved against the trend and the zen guy you spoke to didnt mention profiles you then jump to a conclusion I am talking trash. You seem to swing from one fence to another at a moments notice or maybe you just enjoy disagreeing with me.
These profiles are not dynamic, they determine the way DLM behaves, it doesnt switch between these profiles at will. If someone is on a stable profile it is because a line order has been put in to put them on a stable profile, and they will permanently be on a stable profile until a new order goes in to change it. (source senior tech at a isp cant disclose).
-
Ive just seen something that would imply that on both adsl + fttc the ISP has options to order different base DLM profiles too. It also appears that Zen may not be provisioning fttc for speed see this post (http://forums.thinkbroadband.com/zen/t/4252075-re-ip-profile-reset-on-fttc.html)
BT Wholesale still use the profile names of ADSL - so when we order 'Standard' and BT Wholesale say it's standard what they pass on to Openreach is an order for 'Speed'.
With our GEA FTTC orders we order 'Standard' - so that will be the middle (balanced) profile, rather than 'Speed'..
So therefore the 3 base profiles are:-
BTw "Standard" = BToR "Speed"
BTw "Stable" = BToR "Standard"
BTw "Superstable" = BToR "Stable"
and this now (http://forums.thinkbroadband.com/zen/t/4252187-re-ip-profile-reset-on-fttc.html) confirms Zen arent using fttc speed profile like they previously thought they were.... simply because of the confusion in the differing base profile names
Yes, as GEA FTTC is a direct order with Openreach we order the DLM profile using Openreach's names.
They should be changeable in-contract without resetting the term (unlike a shift from standard to elevated traffic profiles, and vice-versa). We've not tried yet though - as the understanding was that we were already using 'Speed' - but we've found that isn't the case.
So are all ISPs using the "Standard" base DLM profile for FTTC? Have other ISP's been caught out with the difference in base DLM profile names thinking that Standard was Speed?
What difference would this make in reality to the EU?
-
The speed profile is often referred as the the "Gaming" profile by Sky at least, aka enables FastPath.
-
The speed profile is often referred as the the "Gaming" profile by Sky at least, aka enables FastPath.
all the profiles enable fast path at start.
-
Ive just seen something that would imply that on both adsl + fttc the ISP has options to order different base DLM profiles too. It also appears that Zen may not be provisioning fttc for speed see this post (http://forums.thinkbroadband.com/zen/t/4252075-re-ip-profile-reset-on-fttc.html)
BT Wholesale still use the profile names of ADSL - so when we order 'Standard' and BT Wholesale say it's standard what they pass on to Openreach is an order for 'Speed'.
With our GEA FTTC orders we order 'Standard' - so that will be the middle (balanced) profile, rather than 'Speed'..
So therefore the 3 base profiles are:-
BTw "Standard" = BToR "Speed"
BTw "Stable" = BToR "Standard"
BTw "Superstable" = BToR "Stable"
and this now (http://forums.thinkbroadband.com/zen/t/4252187-re-ip-profile-reset-on-fttc.html) confirms Zen arent using fttc speed profile like they previously thought they were.... simply because of the confusion in the differing base profile names
Yes, as GEA FTTC is a direct order with Openreach we order the DLM profile using Openreach's names.
They should be changeable in-contract without resetting the term (unlike a shift from standard to elevated traffic profiles, and vice-versa). We've not tried yet though - as the understanding was that we were already using 'Speed' - but we've found that isn't the case.
So are all ISPs using the "Standard" base DLM profile for FTTC? Have other ISP's been caught out with the difference in base DLM profile names thinking that Standard was Speed?
What difference would this make in reality to the EU?
kitz I got no idea which profiles are been ordered by which isp's I am assuming its the standard profile. But thats just a guess.
his post here implies zen customers are ending up on the openreach speed profile.
http://forums.thinkbroadband.com/zen/t/4252183-re-ip-profile-reset-on-fttc.html
I would think most EU's dont care or notice, they just want a internet service that works. But since this DLM works from the top down, all lines start of as fast as possible, and then DLM only kicks in if there is problems. Are users going to notice the difference between the speed profile and stable profile on a unstable line? probably not, either they will be more concerned about the line itself (if stability is really bad) or they will be moaning to the isp about lost speed (if they a geek who wants every ounce of speed) or most likely I think they wont notice if DLM works, it slows the line down and it becomes stable. Remembering that a FTTC line has to be extremely bad for it to get as slow as an adsl line. I think I read on a openreach page that the middle profile is the default if one is not specified on order.
-
Ive not read the patent in minute detail... but it points to more towards tolerance for the parameters... and what is classed as 'unstable' by the DLM system.
../..
It appears to work almost backwards to how youd expect a straight forward DLM to work - more on this later!
OK we know without doubt there are the 3 DLM base profiles that the ISP can use. We know that the DLM is 'supposed' to set your line profile based on those 3 different profiles.
Aside from the different DLM profile names between BTw and BToR it would appear that these would directly equate to the 3 profiles in the patent.
The patent works differently to other more straight forward DLMs, what the patent doesnt disclose is the profiles, but I think its fair to ascertain that both BToR and BTw will be using the same line management system as these wont/cant be housed in the DSLAM/fttc cab. The DSLAM will simply be sending this information to the line management system. The diagram below was for dslmax, but it will work in a similar way for fttc
(https://forum.kitz.co.uk/proxy.php?request=http%3A%2F%2Fwww.kitz.co.uk%2Fadsl%2Fimages%2Fmaxdsl.gif&hash=f6bd872d6f19c3e6a564bf999077a9b872d22a78)
What is different between BTw and BToR is the actions taken on the line when instability occurs. BTw system will tweak SNRm. BToR passes over this more in favour of interleaving, error correction and banding.
What we have to get our heads around is that it doesnt behave like you'd expect it to.. and those 3 DLM profiles will affect how much the line changes. I dont think that the DLM will automatically switch between those 3 base profiles.. thats up to the ISP to decide. However until someone actually has been on the different base profiles we wont know for sure.
It doesnt go:-
If (ErrSecs == n ) {
If (profile == A ) {Interleave = 'x';}
elseif (profile ==B) {Interleave = 'y';}
elseif (profile ==C) {Interleave = 'z';}
;}
Instead its more like
Calculate_tolerance;{
If (ErrSecs == n && profile ==A ) {calculate quality based on table A;}
elseIf (ErrSecs == n && profile ==B ) {calculate quality based on table B;}
elseIf (ErrSecs == n && profile ==C ) {calculate quality based on table C;}
if (quality == 'verypoor') {increase interleave, reset poorcount;}
elseif (quality == 'poor') {increase poorcount;}
elseif (quality == 'ok') {decrease poorcount, decrease goodcount;}
elseif (quality == 'good') {increase goodcount;}
}
if (poorcount >= $pc) {increase interleave;}
if (goodcount >= $gc) {decrease interleave;}
Note
The above is a simplified pseudo code
We dont know the acceptable trigger for $gc nor $pc
"increase interleave" should be taken to mean increase stability by applying interleaving/increase interleaving depth/increase SNRm / apply banding
We know what those steps are for BTw (I have them somewhere), we dont know what the steps are for BToR fttc.
-
Colin whats this asking for proof business?
It is simply the Scientific Method (http://en.wikipedia.org/wiki/Scientific_method)
-
Downstream Profiles (Mbps)
Rate cap level 36 80 40
Rate cap level 35 74 37
Rate cap level 34 67 33
Rate cap level 33 60 30
Rate cap level 32 54 27
Rate cap level 31 49 25
Rate cap level 30 44 22
Rate cap level 29 40 20
Rate cap level 28 35 17.5
Rate cap level 27 32.4 16.2
Rate cap level 26 30 15
Rate cap level 25 27.4 13.7
Rate cap level 24 25 12.5
Rate cap level 23 22.4 11.2
Rate cap level 22 20 10
Rate cap level 21 18 9
Rate cap level 20 16.5 8.2
Rate cap level 19 15 7.5
Rate cap level 18 13.1 6.5
Rate cap level 17 11.4 5.7
Rate cap level 16 10 5
Rate cap level 15 8.8 4.4
Rate cap level 14 7.9 3.9
Rate cap level 13 7.1 3.5
Rate cap level 12 6.4 3.2
Rate cap level 11 5.8 2.9
Rate cap level 10 5.2 2.6
Rate cap level 9 4.7 2.3
Rate cap level 8 4.2 2.1
Rate cap level 7 3.8 1.9
Rate cap level 6 3.4 1.7
Rate cap level 5 3.1 1.5
Rate cap level 4 2.7 1.3
Rate cap level 3 2.5 1.2
Rate cap level 2 2.2 1.1
Rate cap level 1 2 0.128
Upstream Profiles (Mbps)
Rate cap level 18 20 10
Rate cap level 17 19 10
Rate cap level 16 17 8
Rate cap level 15 15 7.5
Rate cap level 14 12 6
Rate cap level 13 10 5
Rate cap level 12 8.5 4.2
Rate cap level 11 7.2 3.6
Rate cap level 10 6 3
Rate cap level 9 5 2.5
Rate cap level 8 4.3 2.1
Rate cap level 7 3.7 1.8
Rate cap level 6 3.2 1.2
Rate cap level 5 2.7 1.3
Rate cap level 4 2.3 1.1
Rate cap level 3 2 1
Rate cap level 2 1.3 0.5
Rate cap level 1 0.8 0.128
-
Thank you BS.. so now we have the rates at which BToR will decrease the line if the "poor count" threshold has been exceeded.
Since they arent using SNRm to improve stability - instead lowering the line rate which increases the SNRm.
Next question - would they be 'borrowing' the BTw system of SNRm -> interleave -> SNRm -> interleave increments only its cap ->interleave -> cap -> interleave... which would be the logical step.
Still no idea of the poor and good count triggers though. It will probably equate to a number of days.
-
I think its more like.
Interleave -> Interleave -> Cap
although taking colin's case into account it would seem they have made it more complex whereby if they recognise certian 'types' of instability they will cap first instead of interleaving.
-
I think its more like.
Interleave -> Interleave -> Cap
Ive no doubt that it likely will be...in fact it may even be interleave -> interleave -> interleave -> cap in some instances.
This is an area where users observations become useful ???
-
@ Colin S,
Spot on with your assumption !
We might also suggest that the execrable subcontractor practice of commissioning VDSL services without the vital ability to perform full pair quality and VDSL line tests should be ceased immediately as it must surely contravene the sale of goods and services act.
Were that the case, perhaps Walter's wheelbarrow would not burn quite so much rubber !
Kind regards,
Walter
-
@ Colin S,
Spot on with your assumption !
We might also suggest that the execrable subcontractor practice of commissioning VDSL services without the vital ability to perform full pair quality and VDSL line tests should be ceased immediately as it must surely contravene the sale of goods and services act.
Were that the case, perhaps Walter's wheelbarrow would not burn quite so much rubber !
Kind regards,
Walter
I wasnt sure what you meant but looking back I assume you mean this
Have I got that right, and am I right in guessing that your point is that even 'bad' lines (that perhaps might have merited activation in a 'stable' profile) are still activated in an open one?
Personally I find it fair that all accounts are provisioned on an open profile. Yes there may be some instability for a few days whilst the DLM kicks in, but at least it gives a long line the opportunity of getting the best speeds it can in the long term. There may be some longer lines which run pretty stable this way. We all know of the used phrase "Give the line up to 10 days to settle in" really means give the DLM chance to find the best level.
Whilst I agree with you on the use of subcontractors, that is a matter separate from the DLM. Even with a BToR install and all his kit, he would not be able to second guess how the variance of SNRm will affect the line over a period of hours, nvm over a longer period. Things could get even more interesting with the advent of self install FTCC, but again a different topic.
What this thread has disclosed is the DLM being able to identify & mark 'very poor' lines. These are fast tracked through the DLM penalties and therefore they should stabilise much quicker.
The reason I agree with the open profile, is the alternative system is much worse. In the early days of Max dsl, how many good lines got stuck at 2Mb and would have sat there forever not able to reach its full potential. That said my favourite DLM of all time is that of Be* Broadband. It had a KISS approach.. but regretfully I fear that even this may be too complicated for non-technical users. It may also have worked fine on adsl2+ but its doubtful whether it would have been quite so straight forward with all the additional vdsl frequency bands, and they too may have had to come up with an alternative dlm if they ever had gone down the vdsl route.
-
yeah a open profile is better than other types, sky is an example, their DLM starts of slow and works its way up, thats worse than starting fast as possible. Like you said kitz I also remember those 2mbit start profile son adsl max (which often got stuck as well).
-
yeah a open profile is better than other types, sky is an example, their DLM starts of slow and works its way up, thats worse than starting fast as possible. Like you said kitz I also remember those 2mbit start profile son adsl max (which often got stuck as well).
My line was forever getting stuck on 2Mb. I hated it, it happened so many times and would take ages to get reset. Its one of the prime reasons I moved to Be as soon as they came here (annex-m was the other reason) :)
Sky's is just plain silly.. my parents line is stuck at about 1.6Mb when it should be getting about 6-7Mb.
Months later and its still not budged from 1.6Mbps... I still need to sort it and tackle Skys support one day.. because dad cant. The non technical users have a hard time and have to put up with the reboot your router rubbish which of course doesnt make the slightest difference.
-
I swear the DLM hates me, previous sync of 88 days and before that 36 days. Had a lovely power blue transformer explode causing a couple of instance power cuts, DLM now wants to put banding on my line at 20/4.6 with Interleaved off (normal depth 450ish, 3INP). 34 days later and still not back to my old 27/5.2 though for some reason all the daytime RFI/REIN or what ever it was has now cleared, maybe the power cut broke the issue for me? :D
-
I swear the DLM hates me, previous sync of 88 days and before that 36 days. Had a lovely power blue transformer explode causing a couple of instance power cuts, DLM now wants to put banding on my line at 20/4.6 with Interleaved off (normal depth 450ish, 3INP). 34 days later and still not back to my old 27/5.2 though for some reason all the daytime RFI/REIN or what ever it was has now cleared, maybe the power cut broke the issue for me? :D
Similar to me. I think DLM is setup so that if 'instability' is still detected after a number of consecutive 'negative actions' (such as applying higher levels of interleaving) then it bands the connection at a lower speed and disables interleaving altogether, without any future prospect of changing unless reset. Mine must've been at 60/20 with no interleaving for a good number of months now, with the occasional disconnection every 30-45 days when something decides to change the output power.
-
I swear the DLM hates me, previous sync of 88 days and before that 36 days. Had a lovely power blue transformer explode causing a couple of instance power cuts, DLM now wants to put banding on my line at 20/4.6 with Interleaved off (normal depth 450ish, 3INP). 34 days later and still not back to my old 27/5.2 though for some reason all the daytime RFI/REIN or what ever it was has now cleared, maybe the power cut broke the issue for me? :D
Similar to me. I think DLM is setup so that if 'instability' is still detected after a number of consecutive 'negative actions' (such as applying higher levels of interleaving) then it bands the connection at a lower speed and disables interleaving altogether, without any future prospect of changing unless reset. Mine must've been at 60/20 with no interleaving for a good number of months now, with the occasional disconnection every 30-45 days when something decides to change the output power.
Worth noting I had a DLM reset on Monday now on 26.7/5.1 and I'm still on Fast Path!
Was rather funny the CS lady said my line wasn't banded, then she tried telling me the NGA help desk didn't exist. I did have some laughs but god do they wind me up...
-
I swear the DLM hates me, previous sync of 88 days and before that 36 days. Had a lovely power blue transformer explode causing a couple of instance power cuts, DLM now wants to put banding on my line at 20/4.6 with Interleaved off (normal depth 450ish, 3INP). 34 days later and still not back to my old 27/5.2 though for some reason all the daytime RFI/REIN or what ever it was has now cleared, maybe the power cut broke the issue for me? :D
Similar to me. I think DLM is setup so that if 'instability' is still detected after a number of consecutive 'negative actions' (such as applying higher levels of interleaving) then it bands the connection at a lower speed and disables interleaving altogether, without any future prospect of changing unless reset. Mine must've been at 60/20 with no interleaving for a good number of months now, with the occasional disconnection every 30-45 days when something decides to change the output power.
Worth noting I had a DLM reset on Monday now on 26.7/5.1 and I'm still on Fast Path!
Was rather funny the CS lady said my line wasn't banded, then she tried telling me the NGA help desk didn't exist. I did have some laughs but god do they wind me up...
Lucky! I've had no luck getting mine reset, so I've lived with 60/20 for months lol.
-
I swear the DLM hates me, previous sync of 88 days and before that 36 days. Had a lovely power blue transformer explode causing a couple of instance power cuts, DLM now wants to put banding on my line at 20/4.6 with Interleaved off (normal depth 450ish, 3INP). 34 days later and still not back to my old 27/5.2 though for some reason all the daytime RFI/REIN or what ever it was has now cleared, maybe the power cut broke the issue for me? :D
Similar to me. I think DLM is setup so that if 'instability' is still detected after a number of consecutive 'negative actions' (such as applying higher levels of interleaving) then it bands the connection at a lower speed and disables interleaving altogether, without any future prospect of changing unless reset. Mine must've been at 60/20 with no interleaving for a good number of months now, with the occasional disconnection every 30-45 days when something decides to change the output power.
Worth noting I had a DLM reset on Monday now on 26.7/5.1 and I'm still on Fast Path!
Was rather funny the CS lady said my line wasn't banded, then she tried telling me the NGA help desk didn't exist. I did have some laughs but god do they wind me up...
Lucky! I've had no luck getting mine reset, so I've lived with 60/20 for months lol.
Hmm, it's an advantage when you know the majority of the engineers. I told him what needed to be done and he did it, was as simple as that.
-
Wouldnt it be sooooooooo much easier if BT also had a way that the ISPs could request it too.
Perhaps in time BToR will learn that its better to pass this over to the SP too. Its a waste of a BToRs time and resources to send an engineer out just to get this done. In the early days of maxdsl, the ISPs had to battle with BTw to get a DLM reset and so many manhours will have been lost on both sides. Letting the SP request it, simplifies things all round especially when its obvious that its the reason why a line is held short of its capabilities.
-
Kitz it would but I think they have done this to deliberatly have a barrier in place, it looks clear to me openreach want a situation where its fully automated and overides are a complete last resort. I expect their motivation is that a line heavily interleaved is less likely to generate a complaint than a line just been reset but is dropping out due to a fault (which openreach find cheaper to hide via DLM than to fix).
-
DLM wasn't introduced to 'hide' faults. It was implemented to stabilise circuits, with parameters built in such as 'Fault threshold rates', that if breached would entitle the EU/ISP to have a fault built.
If it was given over to the ISP's, then a scenario could develop whereby they themselves just carry out continual resets on a circuit thus masking a potential line fault. The subsequent downward trend in VDSL faults raised against the ISP, would probably be used as a marketing aid as well ??
I agree it isn't perfect, and can be aggressive sometimes, but to suggest it has been designed, or managed in such a way as to hide problems, is akin to a conspiracy theory, IMO.