www Broadband Line Speed Testers

Hi Mike. I use the BT wholesale line tester. I too have an 80 meg service. It appears to be very consistent. 

I wouldn't worry about that size of variation. It could easily be a limitation between the server and your ISPs network. If you use speedtest.net and choose different servers, you will see what I mean.

I notionally have a 200 MBps service and it varies a lot depending on the tester, by at least 20% and often more. The upload is usually the same though (notionally 10 MBps but in practice almost always 12 Mbps).

best

David

I don't trust any of them: The system's non-deterministic, so there's no one single 'correct answer'.

Huge posted:

I don't trust any of them: The system's non-deterministic, so there's no one single 'correct answer'.

Yes I agree with Huge. The only thing I use a speedtester for is occasionally to check that I am getting something reasonable if I'm finding web pages are taking too long to load. ie is the problem in my broadband connection or somewhere deeper in the ips's network..

best

David

I don't trust them either,  thats why I started the thread asking you guys opinions.  We've had posts in the last week or so about peeps with 2Mb/s & those folks have a right to be far more affected/concerned than peeps like me with 80Mb/s,  Line sync speed aside,   the www testers disagree over the latency (ping) test & the variables with that are even worse,  I just tested with BT, Ookla & BB Speed Check,   27ms,  11ms & 112ms ......    

https://fast.com from Netflix is a quick and dirty (only download speed)  tool that works well. It was one of the first HTML 5 testers so works well on both mobile devices and computers. 

I was told that the big sites like ookla are on the ISPs radar so that when you go there, the isp will make sure to give the best result.

Not sure if that is tin hat territory to be honest.

garyi posted:

I was told that the big sites like ookla are on the ISPs radar so that when you go there, the isp will make sure to give the best result. 

Not sure if that is tin hat territory to be honest.

Not what I found yesterday,  it started off with 'Which' sending me a www survey that included a line test,  they gave me the highest downlink,  BT (Wholesale) (my ISP) were next,  then Ookla, followed by three of the others that peddle switching.   But having run the tests again today it's all different.   Might try testing it with some damp seaweed. 

To be honest I would take these testers with a pinch of salt as they are NOT testing line speed... unless you are on a very cheap ISP with a somewhat over contended backhaul, the most telling indicator is the xDSL synchronisation line speed which will be displayed on your modem, or the modem part of your router. This is the real DSL speed up and down irrespective of the vagaries of the internet...and TCP/UDP transfers

Simon-in-Suffolk posted:

To be honest I would take these testers with a pinch of salt as they are NOT testing line speed... unless you are on a very cheap ISP with a somewhat over contended backhaul, the most telling indicator is the xDSL synchronisation line speed which will be displayed on your modem, or the modem part of your router. This is the real DSL speed up and down irrespective of the vagaries of the internet...and TCP/UDP transfers

Doesn't quite work like that for Virgin Media cable modems....!

best

David

David, no indeed hence why I specifically mentioned xDSL to which it does refer to.. 

There appears a chain in the Virgin backhaul that can case contention, and they specifically quote different throughputs at different times of the day perhaps because of this. I will check, but I am not sure how with Virgin you can validate your line speed... their web site only talks about 'internet' speed which has many variables and they actually recommend their customers use a third part broadband internet speed test site, which I think is rather poor as it gives Virgin huge wriggle room. If that is the method they are providing they should provide their speed test site on their network in my opinion..

I use the BT wholesale line tester and Ookla. I  have the virgin media 200 meg service. The results are consistent during the day but in the late afternoon and evenings, the speed drops away and the results ere all over the place.  

I have always used Ookla, works very well a for me. It gives a good enough idea of what's going on. I am lucky to have a Fibre FTTP service, which is  a 1000Mbit  line speed to my ISP with a 50 Mbit throttle on upload and download as I don't want to pay for the full line rate. You have to remember what the tester is measuring, which is the throughput and ping response to a test server located "somewhere". This will be subject to a bottleneck between source and destination and to the vagaries of congestion.  For xDSL the bottleneck  will be the local link  MyTTP service from Gigaclear as well as being very fast has the advantage of being a very simple network with very few devices (hops) between me and the core internet. This gives much less chance for variability to creep in due to congestion. To see how many hops are involved traceroute is a useful programme. Just pick a destination domain e.g. google.com and run a traceroute. If you are on a service such as BT, you may be surprised  how many devices you have to transit just to get out of their network.

As default Ookla will preselect the test server, but you can test to one  anywhere in the world to see the difference in local and remote servers. 

SB for ADSL the congestion is from the exchange DSLAM/MSAN over the ISP back haul.. the local access is non contended.

For VDSL using GEA ('super fast' cabinets) then the congestion can start at the PCP/DSLAM cabinet and over the the multiple fibre links to to the exchange hand over point, and then congestion can occur in the ISP backhaul. Although with GEA (super fast) there is potentially another layer of possible congestion compared to ADSL it will be typically capacity managed to provide the stated bandwidths.

https://www.openreach.co.uk/or...2063350%20111108.pdf