|
|
We're seeing a significant number of posts on the Plusnet community forums where users upgraded to 80/20 from 40/10 are seeing slower actual speeds.
Is this being seen on any other ISPs?
jelv
Plusnet user since November 2001
|
|
|
What were their actual speeds down and up on 40/10? Were they lower than (say) 36/7Mbps, or were they getting maximum?
My broadband basic info/help site - www.robertos.me.uk
Domains,website and mail hosting - Tsohost. Connection - Plusnet Extra Fibre (FTTC). Sync ~ 56.0/13.9Mbps @ 600m.
"Where talent is a dwarf, self-esteem is a giant." - Jean-Antoine Petit-Senn.
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Allergy information: This post was manufactured in an environment where nuts are present. It may include traces of understatement, litotes and humour.
|
|
|
Here's one topic where the 40/10 speed was 37.5/8.3
http://community.plus.net/forum/index.php/topic,1080...
jelv
Plusnet user since November 2001
|
|
Register (or login) on our website and you will not see this ad.
|
|
|
Strange.
My broadband basic info/help site - www.robertos.me.uk
Domains,website and mail hosting - Tsohost. Connection - Plusnet Extra Fibre (FTTC). Sync ~ 56.0/13.9Mbps @ 600m.
"Where talent is a dwarf, self-esteem is a giant." - Jean-Antoine Petit-Senn.
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Allergy information: This post was manufactured in an environment where nuts are present. It may include traces of understatement, litotes and humour.
|
|
|
80/20 may be resulting in more cross talk and as more people order the drop off in speed is more than it used to be with ADSL.
Only way to be sure, is to have before and after bin plots from modems.
There is a possibility that the rush to order the 80Meg means that service is hitting some contention that the old 40Meg did not. Without access to fibre between cabinet and exchange, and the WBC network too it is impossible to say.
|
|
The author of the above post is a thinkbroadband staff member. It may not constitute an official statement on behalf of thinkbroadband.
|
|
|
|
But only on Plusnet?
|
|
|
Is it though?
There was a slow down reported for BT Infinity that Ofcom noticed but written off as a HH3 issue, but that may have masked this issue.
I am not saying this is the reason for sure, but may account for some of it, or there is no single cause.
|
|
The author of the above post is a thinkbroadband staff member. It may not constitute an official statement on behalf of thinkbroadband.
|
|
|
|
|
|
|
80/20 may be resulting in more cross talk and as more people order the drop off in speed is more than it used to be with ADSL.
But anyone on 40/10 is already running on Profile 17a - so the change to 80/20 shouldn't affect much? Unless it increases cross talk.
Don't think you meant to compare 80/20 with ADSL in that sentance.
Only way to be sure, is to have before and after bin plots from modems.
Sadly OR lock the modems 
There is a possibility that the rush to order the 80Meg means that service is hitting some contention that the old 40Meg did not. Without access to fibre between cabinet and exchange, and the WBC network too it is impossible to say.
Yes, wouldn't be the first time an ISP has under estimated the amount of capacity they need from BT's WBC into their network.
James BT Infinity 2 - 19/9/2012 - Install-sync: 52/12 - Test: 50/10 - Est: 44.6/6.5 Mbps
13 years of broadband - ntl:(512k/1M)/BTbusiness(2M)/Metronet(2M)/Bulldog(8M/16M)/BE(16M)/BT FTTC(50M)
|
|
|
so the change to 80/20 shouldn't affect much?
It will increase crosstalk, because the sync is capped, and so at sync time the bit loading is lower. Cue 17MHz pushing more bits into each bin and using more power, so more cross talk.
Not saying it is the total story, but may be a contributor.
On the locking I know, and it is becoming a pain to help understand issues and see through bluster and bluff from those who probably have no idea of the difference between 6dB and 12dB.
|
|
The author of the above post is a thinkbroadband staff member. It may not constitute an official statement on behalf of thinkbroadband.
|
|
|
so the change to 80/20 shouldn't affect much?
It will increase crosstalk, because the sync is capped, and so at sync time the bit loading is lower. Cue 17MHz pushing more bits into each bin and using more power, so more cross talk.
Ahh, yes, hadn't thought the sync would change - cross talk is a pain.
Not saying it is the total story, but may be a contributor.
On the locking I know, and it is becoming a pain to help understand issues and see through bluster and bluff from those who probably have no idea of the difference between 6dB and 12dB.
The locking is a nightmare both for ISPs and for end users - its a shame Openreach didn't get Huawei to produce a very simple status page that the ISP could have talked a user through accessing.
May OR assumed the ISP would have the information on their management system and wouldn't need to ask the user - which is short sighted given the cost pressures ISPs are under. Reminds me of the old "USB frog" idiocity where the USB plug was the official NTE.
James BT Infinity 2 - 19/9/2012 - Install-sync: 52/12 - Test: 50/10 - Est: 44.6/6.5 Mbps
13 years of broadband - ntl:(512k/1M)/BTbusiness(2M)/Metronet(2M)/Bulldog(8M/16M)/BE(16M)/BT FTTC(50M)
Edited by jchamier (Fri 28-Sep-12 18:36:41)
|
|
|
|
|
|
|
so the change to 80/20 shouldn't affect much?
It will increase crosstalk, because the sync is capped, and so at sync time the bit loading is lower. Cue 17MHz pushing more bits into each bin and using more power, so more cross talk.
Ahh, yes, hadn't thought the sync would change - cross talk is a pain.
I've noticed the same thing - that turning on 80/20 makes things worse for some people.
I should be specific here that the observations are that SYNC speeds are worse. While contention & traffic management issues may appear later, for now the issue is strictly about the 17a profile over the copper.
I haven't seen an obvious cause - but that statements above *must* have some impact. Bit loading is altering in some frequencies, and that is having some level of effect.
But crosstalk? Surely that would be a problem YOU experience when someone ELSE upgrades their own line. Why would your own line degrade when you are switched to 80/20 yourself?
For myself, I wonder if it is interplay between upstream & downstream. The biggest impact to most of the affected lines (*) is probably in expanding the upstream, and yet it is the downstream that suffers.
(*) - The kind of line I believe is affected most is the ones with service around the 40Mbps level, who can expand into the 50's, but actually drop to the 30's.
Whatever happens, it usually seems to trigger DLM into placing profile restrictions, and activating interleaving that was never needed before.
Not saying it is the total story, but may be a contributor.
On the locking I know, and it is becoming a pain to help understand issues and see through bluster and bluff from those who probably have no idea of the difference between 6dB and 12dB.
The locking is a nightmare both for ISPs and for end users - its a shame Openreach didn't get Huawei to produce a very simple status page that the ISP could have talked a user through accessing.
May OR assumed the ISP would have the information on their management system and wouldn't need to ask the user - which is short sighted given the cost pressures ISPs are under. Reminds me of the old "USB frog" idiocity where the USB plug was the official NTE.
On the other hand, many people don't understand the stats being presented, and use them as a whip to the ISP. Without the stats, there is less to use against the ISP. Perhaps they were hoping that the customers could be kept more docile?
It is probably a case of "damned if you do, damned if you don't".
Oh - and Huawei's status page, once unlocked, is wrong anyway - so would only have confused things! The error counts seem to come from the wrong stats.
|
|
|
I should be specific here that the observations are that SYNC speeds are worse. While contention & traffic management issues may appear later, for now the issue is strictly about the 17a profile over the copper. But 17a was rolled out last year?
|
|
|
|
I remember on my old adsl line adsl2+ synced massively lower than adsl2 and a few on here couldnt understand why.
My guess is the extra US sync is causing issues on the DS sync. Of course extra cross talk may also be an issue.
|
|
|
I should be specific here that the observations are that SYNC speeds are worse. While contention & traffic management issues may appear later, for now the issue is strictly about the 17a profile over the copper. But 17a was rolled out last year?
Yes - and I saw changes on my 40/10 package at that time. The shift in some of the bands gave a slightly bigger allocation to downstream at lower frequencies, and my line squeezed an extra 3Mbps from it.
However, everyone was still constrained to 40/10, so the bit-loading that the modems determined during sync was lower than now.
The observations NOW are that it is the change of constraint from 40/10 to 80/20 that does the damage, not the introduction of the 17a profile alone.
|
|
|
|
What you need to demonstrate that, is to have the stats from someone who is on 40/10 now and compare them with stats after they move to 80/20, especially if their sync has actually fallen after the move.
|
|
|
Which, as BT have locked the modems, is quite tricky 
|
|
|
FWIW, The switch from 40/2 to 40/10 also has a negative effect upon DS sync speed for those connections that are unable to max out the 40Mb cap.
Plusnet & I experimented with 40/2, 40/10 & 80/20 on my "up to 1000m" connection.
These are typical results
| Text | 1
23
4 | Service DS sync speed
40/2 up to 35Mb40/10 up to 30Mb
80/20 up to 30Mb |
Attainable DS Rate has been around 34Mb to 36Mb for all 3 services.
US sync speeds increased from 2Mb (capped) to around 5Mb on both the 40/10 & 80/20 services on my longer length connection.
So, for my connection, a 3Mb increase in US sync speed causes a 5Mb decrease in DS syncs speed.
A few other sub-40Mb users (not only Plusnet) have also mentioned the same effect on switching to a higher US service.
|
|
|
|
But 40/10 to 80/20 had no effect?
|
|
|
|
Interesting. I didn't realise you'd been testing quite that much.
Some questions...
- Did they fix your problem in the end? Was it physical?
- Did you do these tests after the fault was fixed, or as part of the investigation?
- Do you have graphs to show the different bit-loading patterns?
It certainly suggests that there is a case for allowing people to self-cap the upstream, in order to improve the downstream.
|
|
|
But 40/10 to 80/20 had no effect?
That's right.
My connection is unable to use any of the higher frequencies needed for up to 80/20, so it was already using everything it could on 40/10.
Therefore no difference was seen by attempting 80/20.
|
|
|
Interesting. I didn't realise you'd been testing quite that much.
Some questions...
- Did they fix your problem in the end? Was it physical?
Yes. Fluctuating SNRM & disconnections from using the phone turned out to be a defective faceplate filter & all the other random disconnections (especially during warm & dry weather) turned out to be a poor underground cable to pole top DP joint.
The DP joint actually fell apart as soon as the engineer touched it
- Did you do these tests after the fault was fixed, or as part of the investigation?
SOME testing was done during the investigation (mainly as Plusnet's experimental attempt to induce a DLM reset without engineer intervention - it didn't reset DLM).
The same pattern was seen, but as my connection was so unstable at that time, I discount the data as unreliable.
- Do you have graphs to show the different bit-loading patterns?
I have so many graphs, without literally spending hours on it, it would be difficult to pick out the most relevant.
Here are a couple though:-
2Mb Capped US
Up to 10Mb US
The differences are slight, but visible.
Not really relevant to this topic, but the change from 8c to 17a also had some effect on lower speed connections.
Mine, for example, could use all the 8a DS frequencies (up to tone 1627).
However, it cannot use all the 17a DS frequencies due to attenuation over distance from the cab.
The highest tone it can use is approx 1750 (approx 7.5MHz).
Some connections that could use slightly higher frequencies up to say tone 1963 benefited more from the switch than others.
Mine only benefited slightly from the increase from DS tone 687 to 859 & lost a little as the next tone band now also starts at a slightly higher frequency.
8c DS tones - (32,687) (1192,1627)
17a DS tones - (32-859) (1216-1963) (2792-3959)
THose are from a Huawei DSLAM. ECI DSLAMS are slightly different on 17a & I have no idea what they were on 8c.
It certainly suggests that there is a case for allowing people to self-cap the upstream, in order to improve the downstream.
The difference isn't great, but a slightly higher US speed does more than offset the lowered DS speed for my own useage - working from home quite a lot & needing to upload data to the office servers.
EDIT:
Hmmm. I wonder what effects would be seen for someone who can't max out 80/20 experimenting with 80/10 and/or 80/2.
Any takers?
Edited by deleted (Sun 30-Sep-12 09:08:53)
|
Pages in this thread: 
Print Thread
jelv
