Theorectical vs. Actual Speed

I have recently run the BT Speed Test (www.speedtester.bt.com) and am a little confused by the results.

It tells me that the "Configured download speed for this service is 1000K" with an actual download speed of 744K.

That's all well and good but it tells me parameters of my line are 1344 Kbps download and 448 Kbps upload.

I am unsure how the numbers relate to each other, e.g. 1344 vs 1000 vs 744.

Can anyone explain what is actually happening and is there anything I can do to get the numbers closer to the indicated 1344?

Thanks.

1344 is the speed the modem connects to the telephone exchange.

Data travels in an ATM formet over this link, and the Internet uses IP format. There is an overhead to this, i.e. around 10 to 13% of the connection is lost in the conversion, so the 1344 becomes around 1189Kbps.

Then due to the way BT Wholesale runs it systems, they have a fixed number of speed profiles, the closest to 1189 is the 1000K one that you see displayed, thus they limit your IP (basically what speedtests report) to 1000Kbps.

In essence you are getting 74% of the speed from the line that it will allow you to have.

Thanks for the clear explanation. It makes much more sense now.

Out of interest what is the next speed profile above the one I have?

Full list and the connection speed needed for them at http://www.thinkbroadband.com/faq/sections/radsl.htm...

It is 1250Kbps, and a sync of 1440Lbps

By some twist of fate my sync speed is now 1472 kbps which should put me just over the threshold for a higher line profile speed. Any ideas how I can get this changed?

I also wonder whether there is some other latency issues with my line. I tried pinging www.demon.net and got 0.0% packet loss but it seemed very slow, e.g.

round-trip min/avg/max/stddev = 117.353/133.516/143.996/7.349 ms

Is this normal on this sort of line?

The IP Profile change is automatic and takes between an hour and five days after having a sync that means a higher value applies.

Lower values happen within minutes.