Is The DLM Algorithm Becoming Less Harsh?

Hi all,

Yesterday, I was doing a lot of testing for the reasons in my FEC thread down in the technical section. I must've done at least 10 resyncs, so I expected today that the line would be banded, but no only the DS SNRM was increased from 3 to 4 dB and G.INP was re-enabled on the US.

I thought I'd share this, as if you're planning some work/testing that may involve resyncing the modem several times, you may not have to be as concerned about the result of the actions now...

Edited by deleted (Thu 07-Sep-17 13:58:23)

We�ve not been told it has changed, no.

I'm not aware that you can be told it has changed, as it stays within the realms of Openreach. It was just something I noticed, so wondered if anyone else had experienced this.

In reply to a post by WilliamGrimsley:

I'm not aware that you can be told it has changed, as it stays within the realms of Openreach. It was just something I noticed, so wondered if anyone else had experienced this.

It might depend on how long the modem was disconnected for.
Or as you only did the tests yesterday , the DLM might do the banding etc later.?
But I do not think that OR give out details relating to the DLM system unless they have to.

DLM would've banded the line 1st if it detected "severe" instability. It wouldn't of done the changes this morning, if it thought it wasn't enough.

In reply to a post by WilliamGrimsley:

DLM would've banded the line 1st if it detected "severe" instability. It wouldn't of done the changes this morning, if it thought it wasn't enough.

So it looks like you are correct in posting that the DLM is less harsh.
Or you just did not have the modem disconnected long enough at each re-sync for DLM to take any action.?
.

Ah, I'm talking only a few minutes at most between power down and power up. I think if it's longer than 15 minutes and/or longer than a 15 minute period DLM monitors, it doesn't count it as a resync. I could've done that, but the testing would've taken a lot longer!

You may just have been lucky. A single instance of DLM not intervening is not proof and as OR have made no announcements then we can't answer the question. We could all of course test it but that might result in a lot of people unhappy if DLM does kick in. So, unless someone has insider knowledge we probably won't know why yours survived the testing unscathed.

Yes, I agree there, Ian. The last thing I want is several people doing what I did, but that if similar actions are necessary, then they may also get away with it. DLM is an interesting topic, just a shame that we can't access the "behind the scenes" workings of it.

In reply to a post by WilliamGrimsley:

I'm not aware that you can be told it has changed, as it stays within the realms of Openreach. It was just something I noticed, so wondered if anyone else had experienced this.

You wouldn�t be told, but I�d like to think I would be told given I work for them and what my job is!

Haha, but would you be allowed to give out that information on a forum? Probably not.

But if that was the case then Icarus would have likely not posted at all rather than posting to say they hadn't been told anything. Not being allowed to say is very different from specifically saying there is nothing to say

That's a very good point.

Tracking DLM has always needed masses of evidence gathered over time.

Nowadays we get less evidence reported by people, often with less statistical detail. And what we do see is less consistent. Far less consistent.

It is proving to be a slippery customer these days and, frankly, not worth the effort.

But, given the potential pain of getting capped, and having experienced that pain first hand yourself, I'm surprised you aren't giving it the respect & patience it demands (if not deserves).

In reply to a post by WilliamGrimsley:

Haha, but would you be allowed to give out that information on a forum? Probably not.

Then again, that type of information would usually be shared by Openreach with CPs. At which point really it�s public information anyway. I�ve never found Openreach to be particularly secretive with things like that, as an open network provider it wouldn�t really make much sense to be.

It would be very different from me looking on the system, seeing a planned route for FTTP related fibre (which I have seen) and then blurting it out on here. As that might give Virgin Media or whoever else a head start.

To add if we start to see an obvious trend in the speeds of VDSL2 customers and its not obviously linked to a previous announcement e.g. G.INP, low SNR margin roll-outs, vectoring then we'd chase for an official response.

At the end of the stat the VDSL2 stats tracking can be interesting, but the only thing that matters to the majority is what speeds they get and what that allows them to now do.

In reply to a post by WWWombat:

But, given the potential pain of getting capped, and having experienced that pain first hand yourself, I'm surprised you aren't giving it the respect & patience it demands (if not deserves).

I'm struggling to work this out. Are you saying that I'm supposed to be giving DLM respect? Because at the end of the day, I wish I could, but I can't just yet. DLM is a great piece of engineering, but like I and so many others have seen/experienced themselves, it's very tempramental.

Edited by deleted (Fri 08-Sep-17 12:11:07)

In reply to a post by WilliamGrimsley:

I'm struggling to work this out. Are you saying that I'm supposed to be giving DLM respect?

I'm saying that you do need to treat the DLM with respect. That doesn't mean it deserves that respect out of being good, but out of potentially being very very bad.

If you woke up tomorrow morning to find a cobra staring at you, you would treat it with utmost respect ... out of fear.

Why poke a snake when you don't have to?

Edited by deleted (Fri 08-Sep-17 13:38:52)

LOL, so basically you're saying that due to the resyncs I caused on Wednesday, I should respect DLM. DLM should make it's decisions based on the line's condition, not the person causing the problem. That's an interesting opinion... I certainly don't think you can compare DLM to a snake.

Also, you're infering that I caused those resyncs to try and get banded again, not sure where you got that from. I couldn't avoid causing the resyncs because of the type of testing I was doing. If you read this thread, you'll understand why.

Edited by deleted (Fri 08-Sep-17 16:12:36)

In reply to a post by Icaras:

In reply to a post by WilliamGrimsley:

I'm not aware that you can be told it has changed, as it stays within the realms of Openreach. It was just something I noticed, so wondered if anyone else had experienced this.

You wouldn�t be told, but I�d like to think I would be told given I work for them and what my job is!

True Icaras , but I do not see why OR keeps the workings of the DLM so secret as it is very important to the user to understand what they can and cannot do.
Sometimes even on a very stable line it is important [if you build PC`s] to be able to disconnect the modem without problems from the DLM system.?

In reply to a post by ironman12345:

Sometimes even on a very stable line it is important [if you build PC`s] to be able to disconnect the modem without problems from the DLM system.?

I don't quite see what you mean there.

Just don't have the PC connected to the modem (or modem/router)? If it is important to test LAN connectivity, then the test rig should have a separate modem and router.

Edited by RobertoS (Sat 09-Sep-17 22:52:07)

In reply to a post by RobertoS:

In reply to a post by ironman12345:

Sometimes even on a very stable line it is important [if you build PC`s] to be able to disconnect the modem without problems from the DLM system.?

I don't quite see what you mean there.

Just don't have the PC connected to the modem (or modem/router)? If it is important to test LAN connectivity, then the test rig should have a separate modem and router.

Yeah I agree RobertoS , Was not thinking straight when I posted that.
But it does not alter the fact that the user needs a lot more info about the DLM from OR.

In reply to a post by ironman12345:

But it does not alter the fact that the user needs would like a lot more info about the DLM from OR.

Corrected that for you.

BT/OR are more interested in minimising connection problems for themselves than satisfying customer curiosity, and if that means a few nerds (and I include myself in that!) don't get everything they would like then tough.

Basically- if you don't like it, get yourself a leased line.

Edited by billford (Sun 10-Sep-17 00:02:40)

And with a leased line the termination kit is often managed and thus closed off to you...so obsessing about light levels is not possible.

I think you have the wrong of the stick...not the first time.

The DLM knows nothing about 'you the person', but if you are going to attempt to laboratory test the line to try and discover why errors that are being corrected are appearing then you do very much risk confusing the DLM system, i.e. it will have been written with the average user in mind, not those attempting to run a DSL diagnostics facility.

If you want to fund the cause of the FEC you ideally need to find a modem with microcode that will log the all the conditions at the nano second of each FEC i.e. why try to understand why it was an error and why it was able to be corrected. You also need access to the DSLAM logging as its decision making can have an impact on how things behave, plus a full map of the cross talk environment for your bundle, which DSLAM records will give you.

In reply to a post by ironman12345:

True Icaras , but I do not see why OR keeps the workings of the DLM so secret as it is very important to the user to understand what they can and cannot do.

I think it would be a tiny minority of users that would be worried by this. The OP has run up against DLM, voluntarily, in the past but very few would be messing with it.

In reply to a post by MrSaffron:

And with a leased line the termination kit is often managed and thus closed off to you...so obsessing about light levels is not possible.

Nearly always in my experience. Laser terminates on an operator-provided NTE and you get an Ethernet hand-off.

Saves headaches with end users using the wrong optics.

In reply to a post by Ignitionnet:

In reply to a post by ironman12345:

True Icaras , but I do not see why OR keeps the workings of the DLM so secret as it is very important to the user to understand what they can and cannot do.

I think it would be a tiny minority of users that would be worried by this. The OP has run up against DLM, voluntarily, in the past but very few would be messing with it.

I understand thanks! So the answer to the DLM , is if you cannot control it do not mess with it.
But I had problems about a year ago with the DLM knocking about 20Mbps off of my line sync and then took 4 months to recover it. Since then I have respect for the DLM.

I don't know about respect but you could be right. For some reason DLM has been paying a lot of attention to my line over the past month. The speed has gone up, then it went down. This morning saw another reset back up to 71.8mb (from 68mb). Given that in the past year I have had very few and far between resets (I could count them on a couple of fingers), I'm impressed. It's found me at last.

If only the resets were always plain sailing though. For some reason my modem didn't like the one last week..... TBB gave me 3mb download, I stopped the test. Ookla gave me this I ended up switching off the modem and router and restarting, with fingers crossed. Happily it started syncing at a higher rate. DLM must be maturing!

In reply to a post by 69bertie:

I don't know about respect but you could be right. For some reason DLM has been paying a lot of attention to my line over the past month. The speed has gone up, then it went down. This morning saw another reset back up to 71.8mb (from 68mb). Given that in the past year I have had very few and far between resets (I could count them on a couple of fingers), I'm impressed. It's found me at last.

If only the resets were always plain sailing though. For some reason my modem didn't like the one last week..... TBB gave me 3mb download, I stopped the test. Ookla gave me this I ended up switching off the modem and router and restarting, with fingers crossed. Happily it started syncing at a higher rate. DLM must be maturing!

I do not know if OR have revised the DLM system , but personally I think that it has become "less harsh" in the way that it works.?