Do certain ISP's manipulate Speedtests?

[Totempole]

This is something I've experienced over time and across various ISPs (mostly IS based uncapped):

Example: My internet seems sluggish on my uncapped account, so the first thing I do is run a Speedtest. I test both local and international and results seem fine for a 2Mb line, lets say 1650Kbps Download and 450Kbps Upload. The results are good, but my connection still feels sluggish.

I try a test account such as the free one offered by WA, and everything seems far better. At this point, I start to think I'm just imagining things, but when I switch back to my Uncapped account, everything seems to have drastically slowed down again.

Finally I perform Glasnost Test, to test if my ISP is shaping me on HTTP. 500 seconds pass and the test results seem fine at first, saying that my ISP doesn't appear to be shaping me. Upon closer inspection of the results, the first thing I notice is that the max download shows as just 270Kbps and upload at 400Kbps. So I'm not crazy after all, I am indeed being throttled, which is probably why Glasnost test doesn't pick it up as it only tests for shaping.

The point is, if Speedtests aren't being manipulated, then why am I seeing full speed on Speedtests when I am being throttled?

 

[AstroTurf]

what are your ping?

 

[AstroTurf]

Also, they likely do not throttle speedtest.net.

 

[Totempole]

what are your ping?

35-57ms on my Uncapped account
11-13ms on any other account

Also, they likely do not throttle speedtest.net.

This is what I'm thinking. Is it too much to ask for unbiased Speedtest results?


I know that I've been throttled (and with good reason), in fact I intentionally got myself throttled for the purpose of this test, so I'm not complaining about being throttled, just that Speedtest is of no use if it's biased in favour of the ISP.

 

[Chevron]

This is something I've experienced over time and across various ISPs (mostly IS based uncapped):

Example: My internet seems sluggish on my uncapped account, so the first thing I do is run a Speedtest. I test both local and international and results seem fine for a 2Mb line, lets say 1650Kbps Download and 450Kbps Upload. The results are good, but my connection still feels sluggish.

I try a test account such as the free one offered by WA, and everything seems far better. At this point, I start to think I'm just imagining things, but when I switch back to my Uncapped account, everything seems to have drastically slowed down again.

Finally I perform Glasnost Test, to test if my ISP is shaping me on HTTP. 500 seconds pass and the test results seem fine at first, saying that my ISP doesn't appear to be shaping me. Upon closer inspection of the results, the first thing I notice is that the max download shows as just 270Kbps and upload at 400Kbps. So I'm not crazy after all, I am indeed being throttled, which is probably why Glasnost test doesn't pick it up as it only tests for shaping.

The point is, if Speedtests aren't being manipulated, then why am I seeing full speed on Speedtests when I am being throttled?

IS uncapped accounts use a star rating system.

If your stars drop you will be shaped.

 

[mpdjhb]

Have the same issue using testmy.net.

Speedtest.net shows awesome speeds - close to theoretical maximum for my line (local) - while testmy.net shows me at 10% of my line speed, and 80% "below the average".

 

[AstroTurf]

35-57ms on my Uncapped account
11-13ms on any other account



This is what I'm thinking. Is it too much to ask for unbiased Speedtest results?


I know that I've been throttled (and with good reason), in fact I intentionally got myself throttled for the purpose of this test, so I'm not complaining about being throttled, just that Speedtest is of no use if it's biased in favour of the ISP.

Not speedtests fault.
Try one of the other tests (what does Mybroadband's say?)

 

[Nether]

Yip, i'm pretty sure of it, sometimes, when I try to buffer a youtube vid, it would be terribly slow, to remedy this I quickly run a speedtest, which miraculously cures this ailment and the vid start to buffer really well.

Don't have solid prove yet, but I've been surprised by how well it works, going to experiment a bit and record some results.

 

[Necuno]

This is something I've experienced over time and across various ISPs (mostly IS based uncapped):

Example: My internet seems sluggish on my uncapped account, so the first thing I do is run a Speedtest. I test both local and international and results seem fine for a 2Mb line, lets say 1650Kbps Download and 450Kbps Upload. The results are good, but my connection still feels sluggish.

I try a test account such as the free one offered by WA, and everything seems far better. At this point, I start to think I'm just imagining things, but when I switch back to my Uncapped account, everything seems to have drastically slowed down again.

Finally I perform Glasnost Test, to test if my ISP is shaping me on HTTP. 500 seconds pass and the test results seem fine at first, saying that my ISP doesn't appear to be shaping me. Upon closer inspection of the results, the first thing I notice is that the max download shows as just 270Kbps and upload at 400Kbps. So I'm not crazy after all, I am indeed being throttled, which is probably why Glasnost test doesn't pick it up as it only tests for shaping.

The point is, if Speedtests aren't being manipulated, then why am I seeing full speed on Speedtests when I am being throttled?

Apparently openweb does / did

I really got sick of all of OpenWeb`s bull...t

Why fight them? Just leave and find another ISP.

PS. Do a test like this to determine how much Openweb throttles your downloads:

Upload 2 identical files to the same server.
Name the one anything ie. download.zip
Name the second one and add the word "speedtest" to the file ie. download.speedtest.zip

Now test the speeds on both.
The download.speedtest one will break the speed test record whilst the other one will get throttled to death.

I use a premium service for downloading and it`s possible to change the download file`s name before downloading so while I was with OpenWeb I just added .speedtest. to all my download files and they came down at full speed 24/7

Still left OpenWeb and it`s been my best decision yet concerning ISP`s

 

[gconey]

No certain ISP's don't manipulate speed tests, they all do.

 

[Necuno]

No certain ISP's don't manipulate speed tests, they all do.

It is a bit like 3d futuremark

 

[Totempole]

IS uncapped accounts use a star rating system.

If your stars drop you will be shaped.

Actually if your stars drop you get throttled. Shaping usually happens regardless of stars. For the sake of the test, I was down to 3 stars. Going to keep pushing until I get to 2 and test again.

Not speedtests fault.
Try one of the other tests (what does Mybroadband's say?)

Not blaming Speedtest, blaming the ISPs. Mybroadband Speedtest also shows my full line speed when I being throttled. Will try test.my tomorrow to see if it reflects more accurate results.

Yip, i'm pretty sure of it, sometimes, when I try to buffer a youtube vid, it would be terribly slow, to remedy this I quickly run a speedtest, which miraculously cures this ailment and the vid start to buffer really well.

Don't have solid prove yet, but I've been surprised by how well it works, going to experiment a bit and record some results.

I would be interested to see the results of this.

 

[Thor]

IS uncapped accounts use a star rating system.

If your stars drop you will be shaped.

Read OP's post, shaping/throttle is not the issue, the issue is speedtest.net is not as reliable a diagnostic tool as I once believed it to be.

35-57ms on my Uncapped account
11-13ms on any other account



This is what I'm thinking. Is it too much to ask for unbiased Speedtest results?


I know that I've been throttled (and with good reason), in fact I intentionally got myself throttled for the purpose of this test, so I'm not complaining about being throttled, just that Speedtest is of no use if it's biased in favour of the ISP.

 

[Wasabee!]

Just choose a different speedtest server and watch it sink. Speedtest to the server connect just next to your exchange is lulz but other side of SA...

 

[Chevron]

Read OP's post, shaping/throttle is not the issue, the issue is speedtest.net is not as reliable a diagnostic tool as I once believed it to be.

I see you're struggling with comprehension again.

He already established that all ISPs prioritize speedtest in the first line of his post. And confirmed it again in post of this thread.

Did you fail comprehension tests in school or something?

 

[Chevron]

Shaping usually happens regardless of stars.

Sorry but this is plain wrong.

5 stars is completely unshaped.

 

[Totempole]

Sorry but this is plain wrong.

5 stars is completely unshaped.

That depends on which ISP you're with. With my current ISP P2P is often shaped between 6am and 5pm even if you're on 5 stars.

5 stars means highest priority on the network pool, it doesn't guarantee no shaping.

 

[Chevron]

That depends on which ISP you're with.

All IS based accounts use the same star rating system for uncapped accounts.

With my current ISP P2P is often shaped between 6am and 5pm even if you're on 5 stars.

5 stars means highest priority on the network pool, it doesn't guarantee no shaping.

Actually that exactly what it guarantees.

ii. All packages on five (5) stars and are completely unshaped.

https://www.imaginet.co.za/content/legal/Uncapped-Trial-Terms-and-Conditions.html

All Uncapped ADSL service options are unshaped.

https://gconnect.wirelessg.co.za/uncappedfairusage.aspx

 

[wingnut771]

That depends on which ISP you're with. With my current ISP P2P is often shaped between 6am and 5pm even if you're on 5 stars.

5 stars means highest priority on the network pool, it doesn't guarantee no shaping.

+1

I can confirm when even on 5 stars, one can feel shaping at times, eg: 300kBps on NNTP at 3pm (the time shaping is at it's worst from my experience) instead of my full 420kBps.

Without knowing the status/load of the network in that point in time, stars are meaningless, as 1 star can feel like 5, and 5 can feel like 1 (exaggerated for purposes of this discussion) depending on load.

 

[Wyzak]

Definitely yes.