Hmmmmmm

Hmm, I seem to have hit a wall now. I have tweaked settings, I have changed routers and filters. Router stats are solid at:
Downstream Upstream
17112 Kbps 1019 Kbps
Line Attenuation 23.4 dB 11.5 dB
Noise Margin 5.4 dB 11.5 dB

and yet my speed test here gives 13.5Mb/s and on speedtest.net is closer to 12.

I have tried a variety of MTU settings. I'm not unhappy with the speeds, I would just like to get the maximum out of the connection. Now I know I will not get a throughput of 17Mb but would be nice to get 15 or 16 if it can be done??

Have you tried the MTU and RWIN?

eta; oops.

Edited by deleted (Wed 23-Mar-11 14:03:41)

What speeds do you get when downloading a test file? speed test sites are not always accurate,use a download manager if you have one ,as they may rate limit each stream, as such use of a download manager would overcome this by connecting multiple streams

Try this in-depth speedtest, you should get at least 15 meg on your sync. Don't fill in any fields and you'll have to do the test twice as your speed is >10 meg. Do post your results!
http://loki09.mpi-sws.mpg.de/bb/bb.php

Downstream Bandwidth: 15112 Kbps Downstream Queue: 14 ms
Upstream Bandwidth: 901 Kbps Upstream Queue: 271 ms

Right. So it would appear, according to this test, I am doing pretty well!

Also, just did a TBB test file download. In Chrome it was showing as 1600+ kBs. I suspect I am not going to get any more than that, but am confident all is ok

Cheers guys

In reply to a post by SimonCandest:

Downstream Bandwidth: 15112 Kbps Downstream Queue: 14 ms
Upstream Bandwidth: 901 Kbps Upstream Queue: 271 ms

Right. So it would appear, according to this test, I am doing pretty well!

Also, just did a TBB test file download. In Chrome it was showing as 1600+ kBs. I suspect I am not going to get any more than that, but am confident all is ok

Cheers guys

I wouldn't pay too much attention to the TBB speedtester. For some unknown reason it doesn't give the correct speed for Talktalk users as proven here:
http://www.talktalkmembers.com/forums/showthread.php...

If you have to use speedtest.net, then use the London server as that is the one with the most capacity and gives the most realistic results

Shh TalkTalk might have to tell you about the basic HTTP throttling that goes on, i.e. at peak times our tester is slower due to it using a single thread. Off people same PC and some link better speeds.

I'd have assumed people would have figured this out for themselves, as it is easy to figure out TalkTalk links

the automated testing inside tbbmeter now uses a mixture of single and multiple thread tests.

In reply to a post by MrSaffron:

Shh TalkTalk might have to tell you about the basic HTTP throttling that goes on,

The Glasnost test can detect that http://www.measurementlab.net/measurement-lab-tools#...

i.e. at peak times our tester is slower due to it using a single thread

Nope, sorry, the TBB speed test gives approximately the same speed whatever time of day or night it is run at. But, since the test files seem to download at the "correct" speed I am not worried.

In reply to a post by MrSaffron:

Shh TalkTalk might have to tell you about the basic HTTP throttling that goes on, i.e. at peak times our tester is slower due to it using a single thread. Off people same PC and some link better speeds.

I'd have assumed people would have figured this out for themselves, as it is easy to figure out TalkTalk links

the automated testing inside tbbmeter now uses a mixture of single and multiple thread tests.

Well i've done all the Glasnost tests and my connection is not being shaped in any way by TalkTalk (shock horror!) and as SimonCandest has pointed out, it makes the no difference if the TBB speedtest is done at 4am on Monday or at 7pm on Sunday. The problem is some TalkTalk users who rely only on the TBB speedtester (ie they think its the bee's knees) to measure their speed will think they're not receiving the correct throughput based on their sync speed and will get on the blower to TalkTalk support....and the poor souls will wrongly blame TalkTalk when the problem clearly lies with TBB speedtester

The problem is some TalkTalk users who rely only on the TBB speedtester (ie they think its the bee's knees) to measure their speed will think they're not receiving the correct throughput based on their sync speed and will get on the blower to TalkTalk support....and the poor souls will wrongly blame TalkTalk when the problem clearly lies with TBB speedtester

I always use http://www.speedtest.net/

Result - http://www.speedtest.net/result/1221195874.png

and the poor souls will wrongly blame TalkTalk when the problem clearly lies with TBB speedtester

Are you so biased you did not see my comment that the speedtest.net result was even slower than the TBB one? Do you not think that kind of puts your argument in a bad light??

SHOCK HORROR REPORT
Speedtest.net shown slower than TBB!!!!!!!

In reply to a post by SimonCandest:

and the poor souls will wrongly blame TalkTalk when the problem clearly lies with TBB speedtester

Are you so biased you did not see my comment that the speedtest.net result was even slower than the TBB one? Do you not think that kind of puts your argument in a bad light??

SHOCK HORROR REPORT
Speedtest.net shown slower than TBB!!!!!!!

Of course on speedtest.net you can get a poor result, it all depends on how congested the server is; but average speedtest.net results ARE higher than TBB tester for talktalk users. Try the London server on speedtest.net, i'm 100% sure it will give you a speedtest result higher than TBB speedtester, ie more closer to your sync speed.
But as this thread on Talktalk forums proves, the TBB speedtest does not give you a realistic result of your throughput....with the exception of 1 user the TBB test is at least 2 meg lower than other speedtests

Edited by deleted (Sat 26-Mar-11 15:55:53)

In reply to a post by baby_frogmella:

In reply to a post by MrSaffron:

Shh TalkTalk might have to tell you about the basic HTTP throttling that goes on, i.e. at peak times our tester is slower due to it using a single thread. Off people same PC and some link better speeds.

I'd have assumed people would have figured this out for themselves, as it is easy to figure out TalkTalk links

the automated testing inside tbbmeter now uses a mixture of single and multiple thread tests.

Well i've done all the Glasnost tests and my connection is not being shaped in any way by TalkTalk (shock horror!) and as SimonCandest has pointed out, it makes the no difference if the TBB speedtest is done at 4am on Monday or at 7pm on Sunday.

So TalkTalk is not throttling HTTP traffic? There must be something wrong with the TBB speedtest then.

I have noticed it has always given me poor results - for years. I suspect it is only good for one or two ISP's who probably prioritise it in their shaping to get good results here.

Will redo my testing at some point

Which was simple....test files during the day when management is thought to be off.

Do the same in the evening, and I found that while during the day on the same PC/link I could saturate link with one HTTP download, at night it took two simultaneous downloads to do the same.

That's a very subjective and not a conclusive test. You would be better off using a repeatable metric such as Glasnost.

I'm not convinced rate-limiting HTTP is a sign of a bad ISP.

Did I say http rate was a sign of rate limiting, just surprised that more people were not aware it was one of the things talktalk uses at times.

Does Glasnost detect the talktalk p2p management?
Oddly Glasnost can upload HTTP data without it being visible to tbbmeter - its downloads are visible and avoid the standard ports.

In reply to a post by MrSaffron:

Did I say http rate was a sign of rate limiting, just surprised that more people were not aware it was one of the things talktalk uses at times.

You said

Shh TalkTalk might have to tell you about the basic HTTP throttling that goes on

What does that mean if not "http rate was a sign of rate limiting"?

Edited by deleted (Mon 28-Mar-11 12:17:50)

Confused now? I hope you are, as you've taken a typo and read a lot into it.
HTTP rate limiting is one item in the traffic management armoury.
If traffic management == bad then its bad, but traffic management is not always bad.

Don't see the sense in wasting an hour or two to look into this now - some people make attempts to be useful and inquisitive not worth the effort.

In reply to a post by MrSaffron:

Confused now? I hope you are, as you've taken a typo and read a lot into it.
HTTP rate limiting is one item in the traffic management armoury.
If traffic management == bad then its bad, but traffic management is not always bad.

Don't see the sense in wasting an hour or two to look into this now - some people make attempts to be useful and inquisitive not worth the effort.

This does not indicate that

Shh