From the explainations ont he test page there are two factors that may at work 1) The program uses UDP rather than TCP to do the test. This eliminates the delays caused by the acknowledgement of TCP packets. Using UDP more accurately represents the maximum bandwidth that the connection can handle and many manufacturers specify their capacities using UDP. The effect of the delay (e.g. the Round Trip Time as shown by a PING) can be significiant on international links. For example, my friends who work for an ISP can get 65 Mbps from the USA if they use UDP which is what the link was rated at or about 2.5 Mbps if they use TCP. If they start multiple TCP links, they all run at about 2.5 Mbps until the total number of links exceeds the 65 Mbps. 2) The test program uses a very small number of packets to do the tests (e.g. 10). If the packet shaping used by the ISP or whoever uses a shaping or speed policing tehcnique such as leaky bucket (read first section of http://qbone.internet2.edu/bb/Bucket.doc if you are not familiar with this method) then the burst size will let such a small number of packets through before the long term (e.g. 1 second) shaping/policing kicks in. This method of testing is great to find out what the line is capable of but doesn't represent what is the fastest speed you can browse etc. at.
Are you subscribed to our RSS feed? You can download the latest headlines and summaries from our stories directly
to your computer or smartphone by using a feed reader.
Trusted
