VDSL Performance & Tuning

Forums › New Zealand Broadband › VDSL Performance & Tuning

1 | ... | 16 | 17 | 18 | 19 | 20 | 21 | 22 | 23 | 24 | 25 | 26 | ... | 285

393 posts

Ultimate Geek

#680123 1-Sep-2012 16:39

SamF:
Summary of Stats
User      DLM CRC
SKYMAX    1    0.06
richms     1    0.14
Allan       1    0.21
SamF      2    0.39
Instinct     2   0.5
jjnz1       2 2

I also noted any trends where users were either swapped to DLM1 or off DLM1 and what their error rates were before this happened:

Swapped onto DLM1
User       CRC
SKYMAX 0.06
SamF      0.07

Swapped off DLM1
User        CRC
maxzzz    1
SamF      12
jjnz1       86

It's interesting that your errors before on DLM2 where quite a lot lower than they are now on DLM2.

In the stats I posted on page 7 I was getting 3 CRC per minute on DLM1, I've since seen it even higher than that (as high as 10 per minute I believe). Despite that high error rate I stayed on DLM1 for a month. Now I am on DLM2 with 0.07 per minute. 0.07 kind of ruins your pattern, but there are probably other factors involved. I only went onto DLM2 5 days ago - we believe DLM will only make one change per week so perhaps I have qualified for DLM1 but haven't been switched yet. It's also possible that since I was on DLM1 before and was downgraded it may take longer to get upgraded or maybe DLM won't switch me back to DLM1 at all.

I don't think you can really compare error rates between DLM levels so probably the most important data is what DLM level and error rate people had before they changed level. But we also need to know how long since the last change.

SamF

1578 posts

Uber Geek

Trusted

#680136 1-Sep-2012 17:27

Yeah, looking back at that, my sync was actually higher too! The main difference I think was that bitswap was on. I wonder if there is a way to force this on?

As for your line, I reckon you were benefiting from DLM's apparent lack of action in the last few weeks. You are right about the error rates only being relevant before the change, that said, there is some relevance to comparing error rates when on DLM-1 as despite there being some lag, you will only normally stay on DLM-1 if your error rate is under the magical threshold.

You are also right that you can't compare error rates between DLM levels directly, however, it still helps us determine the error threshold for changing profiles.

stevehodge

393 posts

Ultimate Geek

#680158 1-Sep-2012 18:08

SamF: As for your line, I reckon you were benefiting from DLM's apparent lack of action in the last few weeks. You are right about the error rates only being relevant before the change, that said, there is some relevance to comparing error rates when on DLM-1 as despite there being some lag, you will only normally stay on DLM-1 if your error rate is under the magical threshold.

I don't think you should be assuming there is only one threshold. I think it's more likely that there is one set of criteria to go from say DLM2 to DLM1 and then looser requirements to stay there. I don't know the details of how DLM works, but if I were designing it I would only look at actual throughput and disconnects for downgrading a connection. It doesn't make sense to move someone from DLM1 to DLM2 if DLM1 is stable and they are getting better speeds regardless of the error rate.

SamF

1578 posts

Uber Geek

Trusted

#680174 1-Sep-2012 18:47

Oh for sure, I agree!

sidefx

3711 posts

Uber Geek

Trusted

#680234 1-Sep-2012 22:06

SamF:
* I realise this seems to be at odds with your experience sidefx, but unfortunately you seem to be the odd one out!

Story of my life :P Looks like my CRCPM have now settled at 0.03 so hopefully I see a change some time soon!

"I was born not knowing and have had only a little time to change that here and there." | Octopus Energy | Sharesies
- Richard Feynman

SamF

1578 posts

Uber Geek

Trusted

#680238 1-Sep-2012 22:26

Yeah, that's a pretty nice error rate so if you don't get changed now I would be VERY surprised!

SamF

1578 posts

Uber Geek

Trusted

#680294 2-Sep-2012 09:45

Hmm, I've just been thinking. My current error rates are:

FECPM: 1400
CRCPM: 0.11

While my CRC rate is good, and within the threshold that we believe is required for upgrading to DLM-1, my FEC error rate is still pretty bad and may well indicate that once interleaving is off my CRC error rates will be significantly higher. Also, looking at my historical error rates; just before I was swapped to DLM-1 the last time, my CRC rate was 0.07 and my FEC rate was 58. Once I was swapped to DLM-1, my CRC error rate went up to 12!

Based on this, I think I'm going to follow sidefx's actions and up my SNRM target to 14 to try to reduce my errors further.

Equinox IT

Cloud spending continues to surge globally, but most organisations haven’t made the changes necessary to maximise the value and cost-efficiency benefits of their cloud investments. Download the whitepaper From Overspend to Advantage now.

sidefx

3711 posts

Uber Geek

Trusted

#680365 2-Sep-2012 13:27

Yup, I think that's a good idea; I figure if the criteria to upgrade DLM are stricter than the criteria that cause it to downgrade, which would sorta make sense and agree with samples in this thread, then it's best to get the MTBE really low so that it upgrades; and then drop the SNRM back down which hopefully won't increase the error rate enough to downgrade it.

Mine dropped further overnight so I'm now sitting on FECPM: 10 and CRCPM: 0.02. Fingers crossed now.

This of course is assuming that changing the SNRM doesn't cause DLM to get "stuck" as a couple of you suspected originally!

"I was born not knowing and have had only a little time to change that here and there." | Octopus Energy | Sharesies
- Richard Feynman

SKYMAX

43 posts

Geek

#680366 2-Sep-2012 13:32

I haven't posted for a while - probably because nothing has changed for me, but here are some of my statistics and settings (which I've never shared before) that I have had for a month now:

I did a manual resync a week ago, as I fitted in a UPS, so had to unplug to get that connected.

-JASON-

sidefx

3711 posts

Uber Geek

Trusted

#680371 2-Sep-2012 13:52

@SKYMAX: Good to know. What made you change the SNRM and RFI settings? And did DLM change for you after modifying them?

"I was born not knowing and have had only a little time to change that here and there." | Octopus Energy | Sharesies
- Richard Feynman

SKYMAX

43 posts

Geek

#680393 2-Sep-2012 14:27

I was DLM-1 before changing the two settings, as I was getting a few too many errors for my liking (around 150 CRC per hour). And I remained on DLM-1 after the changes. I did go from around 38Mbps downstream sync to around 36Mbps, but I'm happy with that since I haven't found anything to saturate the DSL link anyway in terms of content I use.

Note that I have NEVER tried the hack to change the target SNR (margin).

SamF

1578 posts

Uber Geek

Trusted

#680399 2-Sep-2012 14:37

SKYMAX: Note that I have NEVER tried the hack to change the target SNR (margin).

Heh, I hate to break it to you buddy, but changing the 'intended signal to noise ratio' on the 'line settings' page is in fact changing the SNR Margin! :) You increase the SNRM with every notch on that line.

Interesting that a single db increase in SNRM was enough to bring your CRC errors down from 150 to basically nothing! Your line's looking pretty sweet now!

sidefx

3711 posts

Uber Geek

Trusted

#680410 2-Sep-2012 15:05

sidefx: ...get the MTBE really low...

Err, of course I mean really high.

"I was born not knowing and have had only a little time to change that here and there." | Octopus Energy | Sharesies
- Richard Feynman

sidefx

3711 posts

Uber Geek

Trusted

#680411 2-Sep-2012 15:08

SKYMAX: I was DLM-1 before changing the two settings, as I was getting a few too many errors for my liking

So you changed them both at the same time? Just wondering how much difference the RFI setting made vs SNRM.

"I was born not knowing and have had only a little time to change that here and there." | Octopus Energy | Sharesies
- Richard Feynman

SamF

1578 posts

Uber Geek

Trusted

#680417 2-Sep-2012 15:29

Well, it's been 6 hours since I've upped the SNRM Target to 14 and my errors are now:
FEC: 25
CRC: 0.11

Sync dropped by a further 1.7! mbits.

So a big FEC improvement but no change at all on the CRC side! Not sure what that really means tbh. I guess I'm only going to be able to really tell what things look like once I'm on DLM-1

1 | ... | 16 | 17 | 18 | 19 | 20 | 21 | 22 | 23 | 24 | 25 | 26 | ... | 285