UP problem causes ?

Hi,

Just when browsing and using the Xbox the internet often appears to be slow and unresponsive irrespective of what device is being used and/or the router used. - (We use Norton Utilities 16 to keep the PCs clean and running fast).

Just after we moved house about 2 years ago we had FTTC installed we have never been that happy with the line performance. - During one of the Openreach visits we were told by Openreach that there is a small run of aluminium cable in the line between our house and the cabinet and they say that this Aluminium Cable will not be changed as they do not consider that it is am issue, however we do have intermittent noises on the line that we can sometimes hear noises when making voice calls and sometimes the internet performance is dire. - We have tried different routers and that makes no difference.

We like a Fastpath connection, (for obvious reasons) and since we do not get any CRC errors in the DOWN there is no reason to interleave the DOWN but every so often that is what Openreach tend to do when we complain about the generally poor internet performance and even with Fastpath the Thinkbroadband Speed Checker reports just over 30 ms of Bufferbloat.

You can see in the Thinkbroadband Speed Checker graph that the DOWN graph is quite steady and that the UP graph speed is constantly up and down.

http://www.thinkbroadband.com/speedtest/results.html...

When we look at our router stats, (in the ASUS DSL-AC68U), that we use we can see that we presently synced at about 38 mbps in the DOWN and at about 7 mbps in the UP.

There are no CRC errors in the DOWN and we do get quite a few CRC errors in the UP.

The ASUS DSL AC68U Router Log is as follows;
-------------------------------------------------------------------------------------------------------------
System Log - DSL Log



This page shows the detailed system's activities.


DSL Information



DSL Firmware Version
1.0.3.5

DSL Driver Version
FwVer:5.5.2.3_A_A60901 HwVer:T14.F7_0.1

DSL Link Status

up


DSL Uptime
4 days 6 hours 26 minute(s) 4 seconds

DSL modulation

ITU G.993.2(VDSL2), G.998.4(G.INP, Downstream only)


ANNEX Mode

ANNEX A


DSL Exchange (DSLAM)

Broadcom


Current Profile

17a




Line Stats

Downstream Upstream


TCM(Trellis Coded Modulation)

On

On


SNR

5.4 dB

5.9 dB


Line Attenuation

17.4 dB

4.6 dB


Path Mode

FastPath

FastPath


Interleave Depth

1

1


Data Rate

38838 kbps

7023 kbps


MAX Rate

47544 kbps

7023 kbps


POWER

12.4 dbm

2.8 dbm


INP

1.1 symbols

2.3 symbols


CRC

0

184




VDSL Band Status

U0 D1 U1 D2 U2 D3 U3 D4 U4


Line Attenuation (dB)

4.6

17.4

39.1

49.7

59.1

76.2

-

-

-


Signal Attenuation (dB)

4.6

31.3

39.8

49.3

57.7

-

-

-

-


SNR Margin (dB)

5.8

4.9

5.9

5.5

5.9

-

-

-

-

-------------------------------------------------------------------------------------------------------------

The CRC errors only tend to show up in the UP and in percentage terms the UP speed is much lower than the DOWN and I do not understand why since I am supposed to be on an 80/20 profile.

Could this be due to the small rum of Aluminium Cable ?

Since I am supposed to be on an 80/20 profile I cannot understand why the since the line can support at least 38 mbps in the DOWN that I do not get approaching a speed of 20 mbps in the UP or is that unreasonable to expect.

Why is the UP speed only about one fifth of the DOWN speed and why are there errors in the UP ?

Regards,
Fido

It should be highlighted that while upload is 1/4 of the download if you connect at the maximum package speed on the up to 76 Mbps this is not a fixed ratio, and characteristics of the line will cause this 'ratio' to very.

184 CRC errors in 4 days uptime is nothing to be worried about, and as we don't use any artificial smoothing on the speed tests the variation in the upload graph is not indicative of a major problem at one, only if the spikes were dropping to a very low value would I worry.

If the issue is responsiveness when using the internet on multiple devices at the same time, then the issue will depend on what you are actually doing e.g. if uploading a LOT then it is normal for other activities to be impacted.

If large downloads are fine, then issue is more likely to be a DNS issue or even the AV software doing lots of safety checking. Note also it is not 30 ms of bufferbloat, the 30ms refers to your latency, bufferbloat is the two letter grades.

If there are intermittent issues on the line when ALL DSL hardware and microfilters are unplugged then report a voice fault and when engineer attends make sure all DSL kit is unplugged so they cannot blame it.

Hello, Mr Saffron,

Thank you for your reply.

In reply to a post by MrSaffron:

184 CRC errors in 4 days uptime is nothing to be worried about, and as we don't use any artificial smoothing on the speed tests the variation in the upload graph is not indicative of a major problem at one, only if the spikes were dropping to a very low value would I worry.

If the issue is responsiveness when using the internet on multiple devices at the same time, then the issue will depend on what you are actually doing e.g. if uploading a LOT then it is normal for other activities to be impacted.

If large downloads are fine, then issue is more likely to be a DNS issue or even the AV software doing lots of safety checking. Note also it is not 30 ms of bufferbloat, the 30ms refers to your latency, bufferbloat is the two letter grades.

If there are intermittent issues on the line when ALL DSL hardware and microfilters are unplugged then report a voice fault and when engineer attends make sure all DSL kit is unplugged so they cannot blame it.

At this instant of time; the internet seems to be working reasonably well and we cannot hear any problem on voice calls but the voice call noises which are clearly audible are intermittent and are not always there/obvious. - The Openreach engineers never find any problems on any of our equipment or on theirs but sometimes it is like running through treacle when trying to use the internet.

Over the years we have used a few types of AV software and we presently back to Norton which we do not suspect to be an issue and the same issues occurred when we previously used the premium versions of Avira and AVG.

Also, (a few months ago), we carried out a clean install of Windows 10 and only used Windows Defender and that made no difference.

The only reason that we noted the few UP CRC errors and the fluctuating UP speed checker graph was to seek advice if this could be a factor in the poor internet performance we sometime experience.

There could be a DNS problems.

In reply to a post by MrSaffron:

It should be highlighted that while upload is 1/4 of the download if you connect at the maximum package speed on the up to 76 Mbps this is not a fixed ratio, and characteristics of the line will cause this 'ratio' to very.

The UP speed and reported UP errors may not be an issue but it sometimes seems that the UP speed is the forgotten factor in the connection performance equation and I have never understood why the UP speed is normally 1/4, (or in my case always less than 1/5 of the DOWN speed).

If the line is capable for achieving a download speed of 38 mbps then why is the upload speed only 7 mbps, (at best), when the UP profile is supposed to be 20 mbps. ?

Is it that Openreach concentrate mainly on the DOWN speed and they intionally hold back the capacity of the UP or is it that he UP and DOWN speed inversely proportional ?

Regards,
Fido

>If the line is capable for achieving a download speed of 38 mbps then why is the upload speed only 7 mbps, (at >best), when the UP profile is supposed to be 20 mbps. ?

The UP profile is still line length dependent and while most people who get 38 Mbps download speeds can get 10 Mbps or better upload not everyone can due to the way the line behaves at the frequencies used by the upload. This might be the short run of aluminium causing this, or it might be an issue with the line.

>Is it that Openreach concentrate mainly on the DOWN speed and they intionally hold back the capacity of the UP >or is it that he UP and DOWN speed inversely proportional ?

If Openreach was to redefine the VDSL2 standards used on the network more frequency could be allocated to the upload so that people got a more symmetric service, but this would impact the download speeds, i.e. only so much radio spectrum to be shared around.

In reply to a post by MrSaffron:

The UP profile is still line length dependent and while most people who get 38 Mbps download speeds can get 10 Mbps or better upload not everyone can due to the way the line behaves at the frequencies used by the upload. This might be the short run of aluminium causing this, or it might be an issue with the line.

That's exactly the point that I have been making to BT/Openreach but up to now they do not recognise that there is an issue . - However, the fact that the upload speed appears to be capped to, (at best), 7 Mbps indicates that they know that there is a definite problem with either the Openreach line or with their equipment.

How can I persuade Openreach to correct this problem ?

I suspect that a change of ISP provider from BT to Zen would not be of any help as it seems to be an BT/Openreach line/equipment issue but I would consider a change of ISP if that was a factor.

Fido

Until engineers can find an actual voice problem then 7 Mbps upload is not so low that they will be spending time to resolve it.

Frustrating, but there is no legal requirement to provide anything more than 28 Kbps currently.

Note that the UP IP Profile does not behave like the down one. It is nothing at all to do with your connection speed. Nor the length or condition of the line.

It is simply the maximum upstream sync of the Openreach product. 20, 10, or 2.

Edited by RobertoS (Tue 31-May-16 17:35:32)

In reply to a post by Fido:

Is it that Openreach concentrate mainly on the DOWN speed and they intionally hold back the capacity of the UP or is it that he UP and DOWN speed inversely proportional ?

There is only so much spectrum available to VDSL - 17MHz on short lines, but less usable spectrum on longer lines.

Openreach had a choice to make when deploying - did they want speeds to be roughly symmetrical? Or did they want to favour downstream?

The ITU specified a variety of bandplans that can be chosen by the operators to achieve either result, and BT have chosen one that favours downstream speed. The spectrum gets allocated according to the standardised bandplan - and it *roughly* comes out at a 4:1 ratio of downstream:upstream. The actual result depends on the ability of your line - and aluminium will have a small effect on the balance that you see.

One of the main activities performed over FTTC is watching TV - which is fundamentally a downstream operation - but is a key aspect for BT to sell their TV products. It isn't a surprise that BT chose to prefer downstream.

On balance, it is a sensible move. As last reported, the total usage over FTTC (ie gigabytes per month) actually happens at a ratio of 7:1 - the average FTTC line downloads 200GB per month, and uploads 30GB per month.

In reply to a post by Fido:

If the line is capable for achieving a download speed of 38 mbps then why is the upload speed only 7 mbps

An upstream speed of 7Mbps isn't totally out of line with other people getting 38Mbps downstream. When looking at the various lines on MyDslWebStats, I can see other lines that achieve 38-40Mbps downstream tend to achieve 6-10Mbps upstream.

In reply to a post by Fido:

when the UP profile is supposed to be 20 mbps. ?

Are you confusing the "sync speed" with the "IP Profile" here?

For downstream, BT set an "IP Profile" value that depends on the actual sync speed - and asks the ISP to use it to restrict the speed at which data is downloaded to you.

However, the upstream "IP Profile" value has no equivalent role. In the end, it just gets set to match the maximum package speed. If you have bought a package with maximum speeds of 80/20, then the upstream profile is always set to 20Mbps.

Your sync speed, on the other hand, will try to be the best it can be, without going above that maximum package speed. In your case, the best is 7Mbps.

In reply to a post by Fido:

The only reason that we noted the few UP CRC errors and the fluctuating UP speed checker graph was to seek advice if this could be a factor in the poor internet performance we sometime experience.

Your statistics are hard to read, laid out like that, but they really look confused.

- You have an SNRM value of just below 6dB, so your modem has synchronised as fast as it can.

- However, it reports an actual downstream speed of 38Mbps, and an attainable speed of 45Mbps. When we see this, it normally means DLM has intervened to add fairly heavy FEC and interleaving error protection.

- Elsewhere, your INP settings are 1.1 and 2.3 symbols - which are odd values indeed. If DLM hadn't intervened, I'd expect to see zero. If DLM intervened with heavy FEC and interleaving, I'd expect to see value of 3, 3.5 or 4. If DLM had activated G.INP, I'd expect to see a value in the 40's. An inconsistency, at least.

- Elsewhere again, your line reports that G.INP is activated - which normally only comes with very light levels of FEC and interleaving protection ... where we don't see a big difference between actual and attainable speeds. That's an odd inconsistency too.

- And elsewhere again, those stats also report an interleaving depth of 1. On Huawei cabinets, G.INP normally comes with depths of 4, 8 or 16 (ie very light levels). On ECI cabinets, G.INP has been seen with depths of 1 ... but the trial of G.INP on ECI cabinets has been marked with failures, and BT have been taking G.INP away again.

- However, BT hadn't taken G.INP away on all lines - those subscribed to BT TV might have been left alone.

Taken together, these settings are just plain strange. I wonder if your line is on an ECI cabinet, with G.INP still active.

Further, I've seen reports that ASUS modems don't work too well with G.INP on ECI cabinets (though they work OK on Huawei cabinets), so you might be falling foul of this too. However, the regular symptom would be to turn on heavy FEC+interleaving ... and, as I described above, there are inconsistencies visible on that front.

All in all, odd.

Are you on an ECI cabinet?

Are you subscribed to BT TV?

Do you have another modem that you can try? Something other than the ASUS that can provide statistics?

In reply to a post by Fido:

we noted the few UP CRC errors

Incidentally, that number of errors is relatively small. I don't think it is having much of an effect.

In reply to a post by WWWombat:

Are you on an ECI cabinet?

Stats said "DSL Exchange (DSLAM) Broadcom"

So no.

Hi WWWombat,

Thank you for your detailed and knowledgeable reply.

In reply to a post by WWWombat:

All in all, odd.

Are you on an ECI cabinet?

Are you subscribed to BT TV?

Do you have another modem that you can try? Something other than the ASUS that can provide statistics?

Answer = I do not know if the cabinet is or if its not an ECI cabinet.

Answer = I do pay for the BT TV Sports Package and BT HD Sports additional cost but I do not subscribe to BT TV via one of their BT TV boxes.

Answer = At one time we used a Fritzbox Router 7390 and that router registered the same sort of issues in the UP and a similar number of CRC errors in the UP as the Asus Router does but presently my only other router is a BT Homehub 5 which does not, (as I recall provide a lot of statistical information to the user). - Since it was only 5 days ago the BT carried out a profile reset on the line to ensure that it is set to 80/20 I am extremely reluctant to change to another router for testing purposes within the first 10 days since the profile reset as that would upset the BT Line management System which would then drive everything even lower. - Also the wireless performance of the BT Homehub 5 is vastly inferior and our son cannot connect wirelessly in his bedroom.

The reason why the router stats show that G.INP is activated is because I enabled it on the Asus router and as I recall there were a few CRC errors in the DOWN when it was not enabled.

I suspect that I can, (if I want to), reset the SNR Margin from within the Asus Router to an amount of my choice via the Stability Adjustment VDSL option and I am tempted to try this.

I suspect that if I can change the SNR and I dropped it to 3 db the line would still be stable in the DOWN with no errors or drop outs in the DOWN and a possibly a down speed increase to 47 Mbps and that it would still be locked out at 7 Mbps in the UP irrespective of what the DOWN speed was if the DOWN speed significantly increased.

Conversely, if I changed the SNR to 9 db which would reduce the DOWN speed there may also be a small reduction in the UP speed and there could be a latency benefit. - ie. I suspect that the latency may also reduce and I would welcome advice if this potential reduced latency benefit would be the case because I suspect that it is actually the latency issues that I notice most when the internet problems do arise.

Regards,
Fido

In reply to a post by WWWombat:

Incidentally, that number of errors is relatively small. I don't think it is having much of an effect.

In the first 4 days after the profile reset the number of CRC Errors that are reported UP has been small but the issues with the line have not been evident in those 4 days. - When the issues with the line are evident the number of reported CRC errors, in the UP, tend to increase exponentially.