Quick query on line stats after BT visits

Slightly different question and might not be correct to place on the BT board, so if Mods think it's better elsewhere please place this post in the right board section.

So after some advice last time, I grabbed the TP-Link W9970 V2 and replaced the HH5a as the connecting unit to the phone socket, and everything looks hunkydory.

So I just want to confirm that everything is OK after the changes implemented (BT swapped out sections of line at Junction Box 30m away and at Cabinet at around 400m away - by line, not distance, and the change of modem/router).

I believe the following are the most important details from the W9970. Do let me know if more is needed:

Current Rate (Kbps) 19999 76515
Max Rate (Kbps) 29144 78806
SNR Margin (dB) 15.5 6.6
Line Attenuation (dB) 17.7 9.8
Errors (Pkts) 0 0

The speed and Max Rate are similiar to the HH5a before I swapped them out, with the SNR on the HH5a going down to ~5 before I swapped it out, but it retained connection and speed was excellent (and low bufferbloat on the downstream).

With the W9970, the downstream SNR has gone down from 14.5 to 9.8, I believe that is a plus if I'm not mistaken. Whilst the upstream SNR has gone up a bit from 15 odd to 17.7. The Max rate appears to be higher (slightly) compared to the HH5a, so again, a plus, although I have no idea if it's from vectoring, g/inp or anything else that the HH5a wasn't using since the HH5a was rather light on stats and details.

From statgrabber, the following error details were also obtained from the W9970:

power output details I believe:
upstreamPower=66
downstreamPower=126
usNegBandPlanDiscPresentation=3
dsNegBandPlanDiscPresentation=3
SATNus=56,175,278,1023,1023
SATNds=169,205,331,1023,1023
LATNus=56,177,282,1023,1023
LATNds=98,210,330,1023,1023
SNRMarginus=152,162,154,-512,-512
SNRMarginds=66,65,65,-512,-512
transmitPowerUs=-81,-257,64,-1281,-1281
transmitPowerDs=86,78,69,-1281,-1281

uptime and errors:
totalStart=87424
showtimeStart=778597
quarterHourStart=0
X_TP_Bitswap=On
X_TP_SRA=Off
X_TP_AdslModulationCfg=Multimode
X_TP_AnnexType=Annex A/L
X_TP_SupportAdslMode=VDSL2:A;T1.413:A;G.lite:A;G.dmt:A;ADSL2:A,A/L;ADSL2+:A,M,A/L/M;Auto Sync-up:A,A/L,M,A/L/M
[1,0,0,0,0,0]1
receiveBlocks=0
transmitBlocks=0
cellDelin=0
linkRetrain=0
initErrors=0
initTimeouts=0
lossOfFraming=0
erroredSecs=0
X_TP_US_ErroredSecs=0
severelyErroredSecs=0
X_TP_US_SeverelyErroredSecs=0
FECErrors=98398
ATUCFECErrors=23739
HECErrors=0
ATUCHECErrors=0
CRCErrors=12
ATUCCRCErrors=659
ATUCSeverelyErroredSecs=0

I believe my research has yielded that my biggest concern should be Errored Seconds, and a close second being CRC Errors (but mostly CRC occurances over a period of activity).

0 Errored Seconds over 9 days I'm assuming means all should be good.

The 12 CRC Errors in around 9 days I'm assuming is acceptable, and nothing I should be concerned about correct? Although I'm more concened that there was any CRC errors popping up at all, since it all occurred within the last 12 hours (since this post). Should I be concerned at all about this? 12 occuring over last 12 hours?

And should I be concerned about the ATUC CRC Errors? I believe this is on the receiving side of things? So the cabinet or further along the BT network? And nothing to do with this side (W9970) correct?

Oh, one last thing, the Thinkbroadband speedtest shows poor bufferbloat whilst I'm using the W9970 compared to the HH5a, is this because it's more of an entry model device and lacking in specs that would reduce the likelihood of this poor bufferbloat? (HH5a was normally A/C, A/D. W9970 is currenly C/E, sometimes C/F and D/E and D/F) And is there anything I can do to help alter this? If I need to change the modem/router again, I don't mind, it's probably something I'll do later down the line anyway, I just need some good recomendations for a higher spec'd device I guess if that's the only way to deal with the poorer bufferbloat.

Thanks all.

J1NG

Don't worry about the buffer bloat, unless you are someone who uploads terabits of data and need the downstream to perform well, the solution is a router that implements QoS.

Error counts are also in the do not worry column too.

76,515 Kbps connection at 400m is very good.

Did you apply the hack I mentioned to allow the TP-W9970 to be used with DSLstats? So you can tell if G.Inp is enabled?

Thanks for confirming details on the errors and line speeds.

If and when I do upgrade the modem/router again (maybe if I want less plugs being used), I'll look for something that implements QoS as well.

Thanks again MrSaffron.

J1NG

Have not done so so far. Still reading into it and seeing what I need to do specifically to enable the hack (a lot of the info seems a bit scattered on how to get hack running and I haven't read through all of it and understood it yet), don't want to rush anything and do something horribly wrong and end up having to start everything from scratch again. Especially if it ends up causing the W9970 to restart, reboot or shut down. I have horrible experiences with my local DLM, which seems rather trigger happy to cut speeds and take its sweet time about increasing it again.

J1NG

The real cure for bufferbloat is CoDel rather than QOS https://en.wikipedia.org/wiki/CoDel

Another way of detecting whether or not G.INP is active is to obtain your IP Profile from the Further Diagnostics page of the BT Wholesale Performance Test. For this purpose you can ignore the red instructions.

Assuming your connect-time SNR Margin is ~6dB, then without G.INP the IP Profile should be Current Rate x 0.9679. With G.INP active it will be Current Rate x 0.9669.

If your connect-time SNR Margin is any if 5dB, 4dB or 3bB the IP Profile appears not to follow these rules. At least on Plusnet and AAISP. The line speed reported by BT Wholesale and the Openreach GEA test is then well below the Current Rate, and the IP Profile is calculated on the lower figure.

Which rather reduces the speed benefits of the lower margin .

Thanks! That's answered things very quickly and easily.

IP Profile is 73.98 Mbps
I'm connecting at 76.515 Mbps

So G.inp is active on the line as the multiplier is 0.9669 of what I'm connecting at.

J1NG

:: edit ::

Although, I must ask also, the download speed I'm getting from the BT Wholesale Speedchecker is 54.37 Mbps. Far lower than what I should be getting there and far different from what actual (real life download) speeds are. Where does this discrepency come from?

:: edit again ::

Forgot to add the upload speed, which seems more reasonable to what I'm really getting: 17.59 Mbps. So the upload speed seems fine. But on the BT Speedchecker, it looks like I can only hit 54.37 Mbps. Which is a bit weird.

Edited by J1NG (Wed 06-Sep-17 18:13:54)