In reply to a post by MrSaffron:

So if we quadrupled the number of downloads we would like show the same, i.e. DSLreports is a LOT less sensitive to congestion and that is our aim to provide a tester that is close to what you are experiencing.

.

Where does the congestion bottleneck ? is it on my side of the VM network ? or is it outwidth the VM network Thanks i am just trying to get my head round it a bit

If the graphs are exact copies of the ones the speed test showing then shows both speed tests are working correctly in terms of how they handle their maths.

But looking at the graphs for the dslreports it is clear they are reporting not a mean or median speed but something that is like an upper quartile result. We report the median speed, and a burst speed that is the 85th percentile.

As we use less test files our test is more sensitive to congestion and everything even the DSLReports graphs is suggesting congestion. If the DSLreports test just showed the numbers you would have the simple conclusion that everything is perfect, when looking at the graphs (i.e. they seem to understand they are important like we do) you can see something is wrong.

The question is where, and given the wide variety of source files DSLReports uses and they should use different peering links at Virgin Media then points towards local loop issues, or maybe a configuration issue locally not giving enough capacity to new 200 Mbps customers. If our peering link to Virgin Media was congested we would see it affecting all our virgin testers and we are not and we know its not in our network as we can see others with good results too.

[IMG]http://www.speedtest.net/result/4862221431.png[/IMG]

Installed today and it does fly!

Could you do a speedtest on the tbb tester please http://www.thinkbroadband.com/speedtest/speedtest2.html and post the results Thanks

.
.
.
.