Why do I get such vastly differing speed test results?

[Reya]

Three tests, conducted within minutes of each other.

BT speed test:



Think Broadband speed test:



And then, Speedtest.net speed test:



I know there's a difference between MB and Mb (or mb... or whatever ) but - as you can tell - it's always confused me. And I understand that they each test to different servers, etc. But when my experience of my connection (slow speeds in the past couple of weeks, with daily spikes of packet loss that are severe enough to disconnect me from AIM and Skype and make page-loading grind to a halt) is accurate to the Speedtest.net result, why does the BT result say my connection is stupendously good?

[Steve]

Interestingly I'm seeing a much higher latency on the BT Speedtest than with the other 2 almost double in fact, the Speedtest.net grading doesn't that just relate to the speed and latency of your line compared with other UK testers. ie Mines FTTC so it faster and gets a grade A.

[andrue]

Computer networks are a lot like transport networks. Your question is a lot like asking why it takes a different amount of time to get from London to Edinburgh than it does from London to Cardiff.

Speed testers have always been 'unreliable'. Taken on its own a single test is meaningless. Well - it tells you how fast you can download data from - say - the Thinkbroadband servers. But if you think about it you'll realise that you rarely download from those servers so you're testing something you aren't really interested in. It's like testing the journey time from London to Edinburgh when you actually spend most of your time travelling to/from Cardiff.

Where speed testers become useful are primarily when you have a history of results from the same server. By knowing the speeds you typically get you can tell if your connection (or the ISP) are running slower than normal. If you have history from multiple tests you might even be able to spot when a particular route is having problems (ie; the M1 is clogged up but the M4 is still fine). If you test with several testers they ought to be more or less in the ballpark. When you take into account the nature of a residential broadband connection your results are effectively the same - around 9.5Mb/s.

One thing that does stand out though is the TBB graph. It suggests the test was being impacted by other traffic or as you say packet loss. This is what I get:



That's nearly perfect. The variation in the green line suggests some latency somewhere with the download having to slow occasionally to wait for an ACK. The yellow line shows everything screaming along beautifully as multithreading allows one thread to continue while another might be paused.

I'd say your results are okay and the speed 'difference' is unimportant. However the variation on TBB graph is a bit naff. What's your actual sync speed?

[Reya]

However the variation on TBB graph is a bit naff. What's your actual sync speed?

I... have no idea where to find that I'm using a Netgear DG834G and the only reference to speed that I can find is in the router status panel, which states the following:

DownStream Connection Speed    11070 kbps
UpStream Connection Speed    883 kbps

Is that it?

(BTW, thank you for explaining it with the traffic analogy. That made a lot of sense to me. Sometimes I'm left scratching my head and feeling like a bit of a dunce with all the techy stuff!)

[Gary]

Isn't the TBB tester effected by wgat browser you are using? I though they recommend Chrome rather than Safari or Firefox due to flash interaction issues.

[Reya]

I just tried the TBB tester in Chrome and got this:



Out of interest, using Opera about a minute later, I got this:



And here's the same test, again a minute or so later, using Firefox:



Roughly the same speeds as before, and those same jaggedy green and yellow lines (on all three tests/browsers) that Andrue mentioned.

[andrue]

I... have no idea where to find that I'm using a Netgear DG834G and the only reference to speed that I can find is in the router status panel, which states the following:

DownStream Connection Speed    11070 kbps
UpStream Connection Speed    883 kbps

Is that it?

Yup, that's it. The good news is that you seem to be getting the right throughput. There is a roughly 15% overhead with broadband so 11070 kbps should be giving you 9409 kbps..which is 9.4Mb/s.

The graphs you've posted again are interesting. I can't really see what might have that effect but it's better than before. You might want to ask in the TBB forums as there are people there who are better than me at interpreting these things. I don't think you've got any kind of problem though. Just a slightly weird graph and when all's said and done it's only a residential connection. I'd only normally expect a straight graph from a leased line costing a couple of hundred quid a month.

I will say that my graphs are astonishingly good really. IDNet should make them into a poster and put them up in their lobby. I'd like to also thank Acer, DLink, NetGear, Huawei and BT and whatever nameless Chinese company manufactured the switch in my lounge