Quote:
Originally Posted by SyStemDeMoN Also judging by the reply from the OP, he now also knows that speedtest is unreliable, after getting his near top whack with the link I gave him.
Computer experts, BAH ! |
Speedtest is not unreliable, you're misunderstanding what's happening here. When the OP downloads from the Blueyonder service he's simply testing the local connectivity to the Virgin/Blueyonder network. It's like sitting in your lounge and shouting to a friend in the kitchen - of course they're going to hear you, because you're in the same house.
When you use speedtest you're going outside of Virgin's network. This tests "real world conditions", in that it's proving the bandwidth capacity from your machine to a local hop outside of the Virgin network. It's testing Virgin's peering. This is actually the best kind of test you can do - you don't test whether your mobile is working by shouting into the kitchen, you test it by making a call outside your house.
Downloading a file from the blueyonder service is a local test that will basically prove whether you have access to a CIR equivalent to your service. In other words if you're on a 10mb service you'd expect to be able to download from a blueyonder server at 10mbit. If you can't, that's a fault with your service.
However if you use speedtest.net (or download a file from any high bandwidth file site such as oxford.ac FTP etc), you're checking your real world bandwidth usage.
If you find that you get 10mbit downloading from Blueyonder but only 2 or 3mbit downloading from multiple sources external to the Virgin network then you can clearly say that Virgin have a saturation/contention issue. The important bit is the multiple sources - it's possible that speedtest.net, oxford.ac.uk, Microsoft, Akamai, whoever, are experiencing high utilization, or the routes and peers you're using are congested, which is why you need to use a couple of sources.
So make sure you understand what you're actually testing when you test something. Speedtest.net is not "unreliable", you're just doing an experiment when you don't understand what you're testing.