"only for it to degrade to this:

4mbps degradation and you're moaning?

VM users are going from 300 mbps to fewer than 100mbps. Now that's a real drop.

Not the multi thread speed; the single thread speed, which has dropped by almost 40Mbps.

In reply to a post by AdrianPH:

OK. Done some comparative tests and here are the results:

Spedtest.net (multi thread): 74.3Mbps
TBB (multi thread): 69.5Mbps
BT Wholesale test (Multi thread): 65.8Mbps
TTB Ookla Test (multi thread): 65.Mbps
TestMy.Net (single thread): 62.3Mbps
TBB Speed test (single thread): 41.0Mbps

I also downloaded the 1GB test files at:

test.unet.nl/1000mb.bin 53.0Mbps
TBB IGB Test file: 48.5Mbps

Not sure what this tells other than the multi thread tests are generally significantly quicker than the single thread tests, the big exception being tehTestMy.Net results.

The download test files results are quicker than the single thread TBB test results, but nowhere near as quick as the TestMy.Net results.

Those results tell you that your TBB single thread speed test result is suspect and there's nothing wrong with your connection.

Hi,

I monitored and tested my single thread speeds for just shy of 2 months as it was very low. I used TBB. My results showed roughly the following for 80/20 at full sync:

multi thread - 74/76 (barring just before changing)
single thread - 74/76 (between 12am and 3ish pm)
single thread - 6/45 (between 3pm and 12pm)

Now in my case if it was the tester then I don't think it would hit 74+ out of peak hours, it should be the same as the poor results. Same PC, same router, same settings, all the same.

BTW this was something my ISP appeared to accept.

One other bit to remember is throughput guarantees; for example TTB only guarantee 15 throughput (on 80/20). This is something no ISP mentions but can use as a fallback. BTW is higher I believe. Again in my opinion to use sync speeds to secure business is misleading.

Regards
Mac

Edited by deleted (Mon 13-Feb-17 09:34:23)

I'm also loathe to just put it down to the speed tester.

Compare these results with mine taken on the same day. We are both both Pulse8 customers on TTB's network: mine has a higher sync and multi thread performance, but with a significantly lower single thread speed.

To me it's not the TBB speed test that's inconsistent, it's TTB's delivery performance.

Whether it is really worth worrying about is another matter, but that is ultimately up to me to decide.

In reply to a post by jaydub:

I'm also loathe to just put it down to the speed tester.

Then could you explain why your testmy.net single thread speedtests appear to be more in line with others? You've got a difference of ~20mbps between the single thread TBB & testmy.net testers!!! This just about confirms that the TBB single thread tester is giving you wrong results.

Edited by deleted (Mon 13-Feb-17 10:26:56)

In reply to a post by baby_frogmella:

In reply to a post by jaydub:

I'm also loathe to just put it down to the speed tester.

Then could you explain why your testmy.net single thread speedtests appear to be more in line with others? You've got a difference of ~20mbps between the single thread TBB & testmy.net testers!!! This just about confirms that the TBB single thread tester is giving you wrong results.

...Or the TestMy.Net results. Depends where you place your confidence.

If everybody on P8 / TTB were seeing the same I wouldn't keep on pushing this, but nobody to date has come up with a plausible explanation of why some P8 (and by implication TTB) why some users see single thread speeds matching the multi thread speeds and others like me don't.

The disparity appears to be browser and OS independent, so not the same as Adrian has experienced.

Any chance of you publishing your TBB and TestMy.Net results for comparison.

It could also be different network paths or traffic prioritisation. TTB may have specific peering links to wherever testmy.net is or they may be prioritising the protocol used by them thereby hiding issues with congestion.

Consider this....

Are testers reporting, mean, median or as with ookla a slightly skewed mean, drop bottom 30% and top 10%.

The exact same data points can provide very different results.

We use median which on a good connection means effects of av/wifi are minimised, but on a variable single thread test can mean you see a lower figure than some other testers.

Another factor can be where the test is serving its data from, I know about ours, but which locations are people picking for testmynet