Back in June I posted about my Broadband Speeds & Other Mysteries and then upgraded my contract from 7Mbps to 17Mbps.
I have continued to automatically monitor download speed and occasionally done a few manual checks.
It looks like there has been an improvement in service, but not exactly what I was expecting.
I was basically expecting the download speed to increase, but not the variation from measurement to measurement. I guess I thought the peak download speed would rise and the variation would remain much the same.
|Values for last 45 days are for 17Mbps connection. Earlier points are 7Mbps.|
But from my graph it looks like the lowest measured rates are slower than the pre-upgrade lowest speeds!
Admittedly the values on this graph are determined by the SoftLayer (IBM) speed checker (as described here). Whenever I've manually used the BT Wholesale speed checker I usually get a result like this:-
But why trust BTW over SoftLayer? Especially as regular web browsing still seems sluggish to me.
So I'm left with a few questions:-
- is my RaspberryPi based monitor unreliable?
- is SoftLayer a bad speed checker?
- where are the bottle-necks?
- if I were to increase my contract to (say) 30Mbps, would I see a range of results from as high as 30, but still droping down to 5Mbps?
Most of the suggestions for poor internet speed that I've read on the net just don't apply in my case. I get consistent line attenuation and signal to noise figures. My speed is not consistently slow. I check the speed via ethernet.
My next step is to increase the monitoring frequency from 4 hourly to 1 hourly, and to provide some indication of average download speed, maybe on a daily basis.
The search for enlightenment continues!
EDIT: well here's a thing
A bit more research and a little bit of testing and I find that using a UK server, rather than the (presumably) US SoftLayer server, I get more repeatable results. So I'm going to change my command line to:-
wget -O /dev/null wget http://download.thinkbroadband.com/10MB.zip
...and see how it goes.
I suppose I could plot UK speeds alongside international speeds!