Understanding speedtest results

I'm confused.

I finally got my fttc installed yesterday (the BT Infiniy 40/10 package - I'm not a heavy user) and I'm confused by the speed I am getting and the differences in results on the various speed tests.

When it was installed the engineer (technician?) said I was getting the full 40 meg when he plugged his handheld measuring thingy into it.

When I ran the thinkbroadband test I got:

Download 27 Mbps
Upload 4.3 Mbps

so I then ran the speedtester.bt.com one and also for good measure the beta version as well...

speedtester.bt.com
Download speed achieved during the test was - 36.4 Mbps
For your connection, the acceptable range of speeds is 12 Mbps-38.72 Mbps .
Additional Information:
IP Profile for your line is - 38.72 Mbps

Upload speed achieved during the test was - 8.26Mbps
Additional Information:
Upstream Rate IP profile on your line is - 10 Mbps

Beta speedtester
Download speed (Mbps): 40.58
Upload speed (Mbps): 8.09
Ping Latency(ms): 20.00

So, which one represents the speed I am actually getting, and why does the thinkbroadband one think I'm getting 27meg and the others circa 40?

Many thanks in advance.
Richard

Speedtests show what you get to a particular tester. They may go slower than the maximum your line can take for various reasons. (1) things wrong in your local network (eg using a wireless connection). (2) congestion at the exchange or somewhere else along the route to the test server. (3) Inability of the server to go fast enough; maybe because of heavy load. Obviously, (1) will effect all speedtesters, (2) may affect some more than others, and (3) will be specific to each. Thinkbroadband seems to suffer from (3) on download tests for FTTC, and the BT testers suffer badly on upload tests for some reason. (2) and (3) happen much more at peak traffic times; e.g. often early evening.

It is sometimes possible to get high overall speeds my using multiple threads all working in parallel. Some downloaders do this automatically for large files; most speedtesters don't.

There are always slight inaccuracies in tests anyway. These can be made worse by odd buffering effects in local software. For example, the Kaspersky AV used to buffer things up in such a way so that it could check virus as data arrived. This somehow confused some speedtests enough that people often got readings more than 10 times faster than their lines could really manage. The Thinkbroadband speedtest never suffered from this as far as I know.

There are various overheads in transmission that means extra data has to be passed that is not 'user' data: for example every packet needs to have an address to tell it where to go. These overheads account for differences between the 40 on the engineers meter, 38.72 max acceptable speed, and probably some of the 36.4 difference (the rest coming from 1/2/3 above).

Overall, speedtesters are a useful guide, especially if things go really wrong, but don't worry too much about the details.

thanks for the info - i get that they wont all be exactly the same but the difference between 27 and 38 meg is quite a bit and the tests were run within a minute of each other.

Richard

In reply to a post by Anonymous:

When I ran the thinkbroadband test I got:

Download 27 Mbps
Upload 4.3 Mbps

To add to Stephens reply there also appears to be another issue with the TBB speed tester. For some of us the upload figure is badly underestimated. In my case all testers give a download of between 60 to 70Mb/s. All but one give the upload speed as between 14 and 15Mb/s. TBB insists it's only 8Mb/s. I have run test with a friend that proves it's TBB that's wrong.

Example:

But
http://speedtest.net/result/2045114905.png

The difference in download speeds is curious(*) but doesn't surprise me. It's within the ballpark. It's the upload for TBB that is plain wrong for some strange reason. It doesn't appear to be TBB's servers and I know it's not my computer or my connection. I also know that my ISP can support uploads of at least 12Mb/s to my friend on Be's network. It just seems to be the path to TBB that is throttled somehow.

Anyway the real value of speedtests is monitoring how they perform on a reasonably regular and frequent basis. So in my case the fact one speedtester says 70Mb/s and the other says 55Mb/s is unimportant. What matters is if both of them suddenly suffer a haircut. For what it's worth another tester I've started using is http://mcslhr.visualware.com/myspeed/admin.

But really it's not the results that matter - it's the results over time.

(*)Because Speedtest.net has a reputation for over reporting the speed rather than under reporting it.

Edited by Andrue (Thu 05-Jul-12 07:53:10)

In reply to a post by Anonymous:

thanks for the info - i get that they wont all be exactly the same but the difference between 27 and 38 meg is quite a bit and the tests were run within a minute of each other.

Are you using a wired or wireless link? For FTTC it is more important than ever to use wired. FTTC speeds (even 40Mb/s) are pushing wifi to it's practical limit. Yes, it can support that speed but not if you have neighbours vying for space with you. Could the TBB test just have been unlucky? What does a repeat test say?

Edited by Andrue (Thu 05-Jul-12 07:52:44)

In reply to a post by Anonymous:

thanks for the info - i get that they wont all be exactly the same but the difference between 27 and 38 meg is quite a bit and the tests were run within a minute of each other.

Oh I get much bigger differences than that! The TBB speed tester, especially, very rarely gives a true reflection of my speed and has never, not even once since I went onto 80/20, come close to an accurate upload speed. You might find that speedtest.net gives the most consistent results, though you might have to experiment with different servers.

Edit: The other thing to remember is that the TBB tester uses a different definition of Mbps from other speed testers and you need to add just under 5% (actually 4.8576%) to the figure reported by TBB in order to compare it.

Edited by kasg (Thu 05-Jul-12 09:48:08)

What hasn't been said is that on BT Wholesale connections, which you are on, it's the IP Profile that determines the maximum speed, not the raw connection speed. Your IP Profiles show you are indeed connected at the full 40/10, so there is no problem there. (IP profile is determined by the connection speed though).

The results from the BT test that is showing the IP Profiles looks a tiny bit slow downstream, about right upstream. The tbb test is clearly wrong, as the others have explained. There are frequent threads about it in the Talk To The Staff forum, but nothing happens. It isn't helped by the fact some of us get dodgy down, OK up, some the other way round, and some both. Like you. So for them to find the problem is very difficult.

The beta downstream result is impossible, as it is above the connection speed, never mind the IP Profile.

Edited by RobertoS (Thu 05-Jul-12 09:15:32)

In your case, I would suggest the two BT tests are more accurate. However, it is always worth using Windows Task manager - Networking tab, to see what the throughput looks like. Nice and smooth at 3.8% on a Gbit Ethernet card is what to expect. The BT Beta seems to always over estimate a little - maybe 3 to 6%.

Also try, http://mcslhr.visualware.com/myspeed/myspeed_line_ap...

In reply to a post by RobertoS:

The beta downstream result is impossible, as it is above the connection speed, never mind the IP Profile.

It is fairly well known that the BT Beta tends to be between 3 and 6 % higher than actual. I have large numbers of test records where the main BT was preceded by and followed by a BT Beta - in almost every case over reporting.

I have just run a set of tests for comparison.

TBB Java 35.4 Mbps 36298 kbps (37.12Mbps using 1000² rather than 1024²)
BT Beta 39.40 Mbps
BT Main 38 Mbps with 38.72 Mbps sync and an Assured rate or 37.86 Mbps
BT Beta 39.23
TBB Flash 36.47 Mbps or (as above) 38.24 Mbps
Speedtest.net 37.37 Mbps
Visualware 33.2 Mbps (from 33.5 Mbps)

I have an application running which is streaming about 100 kbps (variable) but cannot be interrupted so it will affect all tests equally. However, the results show that the BT Beta overestimates.

Edited by MHC (Thu 05-Jul-12 09:59:08)