Odd speed test results

Just want to ask if all I need to do is disconnect on the router ADSL and then click 'reconnect' ...and shall I leave it until morning?

I had a remote session with the computer supplier this afternoon to fix a few drive space issues, and during that, lost the ADSL connection briefly.
It came back with reasonable settings apart from the downstream SNR, but speed tests are rather varied:

Current ADSL

DMT Status No Defect
Operational Mode ADSL2+
Upstream 1141
Downstream 8668
SNR Margin(Upstream) 6.3
SNR Margin(Downstream) 3.2
Line Attenuation(Upstream) 26.5
Line Attenuation(Downstream) 45.5


BT speedtest:

0 Mbps 7.15 Mbps
Max Achievable Speed

Download speed achieved during the test was - 7.65 Mbps
For your connection, the acceptable range of speeds is 2 Mbps-7.15 Mbps.
IP Profile for your line is - 7.65 Mbps

2. Upstream Test: -provides background information.

Upload Speed
0.97 Mbps

0 Mbps 0.83 Mbps
Max Achievable Speed

Upload speed achieved during the test was - 0.97Mbps
Additional Information:
Upstream Rate IP profile on your line is - 0.83 Mbps

TBB speed test:http://www.thinkbroadband.com/speedtest/results.html...

Speedtest.net gave this result: 2.68 DOWN 0.99 UP

The stats look fine - what do you think is wrong with the downstream noise margin?

The BT speed test is also fine downstream though I'm surprised it is spot on the IP Profile not a bit below. The IP Profile is correct for the sync speed.

What the BT one is talking about re the upstream IP Profile I don't know, but not seen any ADSLx stats for a while. I think it's an arbitrary figure to correspond with the ADSL Max Premium "Max speed 7.15Mbps" at the start. Strange.

The thinkbroadband result is clearly impossible, and points to possible buffering on your system by an Internet Security Suite. Kaspersky used to be the common one but there was another as well.

Speedtest.net, was that speed repeatable?

Which exchange is it please? That might answer a couple of the questions - in particular the stated maximum in both directions on the BT Test.

In reply to a post by RobertoS:

The stats look fine - what do you think is wrong with the downstream noise margin?

It's usually very similar to the upstream one. It's 4.6 down now.

The BT speed test is also fine downstream though I'm surprised it is spot on the IP Profile not a bit below. The IP Profile is correct for the sync speed.

I tried the BT one right after the Speedtest one and was surprised to see it looking 'right'

What the BT one is talking about re the upstream IP Profile I don't know, but not seen any ADSLx stats for a while. I think it's an arbitrary figure to correspond with the ADSL Max Premium "Max speed 7.15Mbps" at the start. Strange.

You remarked on that oddity last time you saw a BT result from me - simple answer is, I don't know.

The thinkbroadband result is clearly impossible, and points to possible buffering on your system by an Internet Security Suite. Kaspersky used to be the common one but there was another as well.

TBB one is usually very similar to both Speedtest and BT.
I've always had Kaspersky, and it's never interfered with anything until recent tests - perhaps an update to it altered things.
This is the latest:
http://www.thinkbroadband.com/speedtest/results.html...

Speedtest.net, was that speed repeatable?

Didn't try that until now - and no, it wasn't! This is the last three results:

2/9/2016 5:26 PM GMT 7.43 Mb/s 0.99 Mb/s

2/9/2016 4:37 PM GMT 2.68 Mb/s 0.99 Mb/s

1/29/2016 12:09 AM GMT 7.47 Mb/s 0.95 Mb/s

Which exchange is it please? That might answer a couple of the questions - in particular the stated maximum in both directions on the BT Test.

NEILB

Looking at underlying data the httpx6 average was around 7.2 to 7.4 Mbps, one of the latent bugs in the old flash, that with some set-ups the timing got mangled so occasionally higher figures than they should be

http://labs.thinkbroadband.com/speedtest should not have this issue.

Classic Kaspersky spike, we did at one time code around how Kaspersky behaved, then they changed their side and caused problems, so in the end tended off those tweaks

A few Zen users have raised concerns over single thread performance with Zen recently.

In reply to a post by MrSaffron:

Looking at underlying data the httpx6 average was around 7.2 to 7.4 Mbps, one of the latent bugs in the old flash, that with some set-ups the timing got mangled so occasionally higher figures than they should be

http://labs.thinkbroadband.com/speedtest should not have this issue.

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

In reply to a post by MrSaffron:

Classic Kaspersky spike, we did at one time code around how Kaspersky behaved, then they changed their side and caused problems, so in the end tended off those tweaks

A few Zen users have raised concerns over single thread performance with Zen recently.

Not sure what 'single thread performance' means to be honest, but I get issues where things don't load as quickly as they should ...think last Thursday or Friday was poor and for a while on Sunday.
Emails can download slowly and images in them take a while to appear, and web pages either tell me there's an error, and then load when re-tried, or take some time to load.

It seems to be OK today.

In reply to a post by TrishaH:

Not sure what 'single thread performance' means to be honest

The TBB tester is relatively unusual in that it performs two different kinds of download tests. The first test is just one download - this is known as a single threaded download. This is what most real-world downloads do. If there are issues somewhere the single thread is blocked and performance drops.

The second test runs six simultaneous downloads - this is known as a multi-threaded download. If there are issues somewhere they often only affect one or two of the threads and the others can often 'pick up the slack' in the meantime.

Under ideal conditions there should be no difference between the two results. However if the multi-threaded result looks good and the single threaded looks bad it typically points to congestion or a fault within the ISP's network. Plusnet suffered this for most of last year and there are suggestions that it was because they outgrew the BT Wholesale product they had been using.

Edited by Andrue (Tue 09-Feb-16 19:36:34)

The exchange didn't answer the questions, as it appears to be standard BT Wholesale backhaul they are using, not their own taken off at the exchange. So I don't see why the BT speed test gives those silly maximum speeds.

The spike in that new tbb test and the general shape also suggests to me Kaspersky type buffering. The only way I know to rule that out, is:-

1) Either disconnect your "session/PPP/broadband" in the router GUI if it has such a button, or if it hasn't then simply disconnect your router from the phone line. Don't do the latter until back in daylight else you are likely to get a slower connection once we've done the following.

2) Disable Kaspersky. IIRC by right-clicking on its icon and getting a popup menu. I find it easiest to disable the whole thing rather than one function.

3) In Windows Security, turn on Windows Firewall.

4) Reconnect by reversing the actions in (1).

5) Run any speed tests you want. Even post the results to here if you wish as this site isn't going to send a virus.

6) Disconnect as in (1) >> Turn off Windows Firewall >> Re-enable Kaspersky >> reconnect.

In reply to a post by Andrue:

Under ideal conditions there should be no difference between the two results. However if the multi-threaded result looks good and the single threaded looks bad it typically points to congestion or a fault within the ISP's network. Plusnet suffered this for most of last year and there are suggestions that it was because they outgrew the BT Wholesale product they had been using.

Thanks for the explanation.

In reply to a post by RobertoS:

The exchange didn't answer the questions, as it appears to be standard BT Wholesale backhaul they are using, not their own taken off at the exchange. So I don't see why the BT speed test gives those silly maximum speeds.

The spike in that new tbb test and the general shape also suggests to me Kaspersky type buffering. The only way I know to rule that out, is:-

1) Either disconnect your "session/PPP/broadband" in the router GUI if it has such a button, or if it hasn't then simply disconnect your router from the phone line. Don't do the latter until back in daylight else you are likely to get a slower connection once we've done the following.

2) Disable Kaspersky. IIRC by right-clicking on its icon and getting a popup menu. I find it easiest to disable the whole thing rather than one function.

3) In Windows Security, turn on Windows Firewall.

4) Reconnect by reversing the actions in (1).

5) Run any speed tests you want. Even post the results to here if you wish as this site isn't going to send a virus.

6) Disconnect as in (1) >> Turn off Windows Firewall >> Re-enable Kaspersky >> reconnect.

OK - thanks!
It seems to be settled and working as it usually does for now ...I'll see how it is tomorrow.
Never had an odd Speedtest before like the first one I got this afternoon!