prioritized traffic (aka 'traffic shaping')

I noted a few months ago that downloads had slowed to around 100KB/s (or slower):

- a 50MB/s cable broadband ISP

- broadband speed test sites show erratic results, but currently thinkbroadband (HTML5) indicates the full 50MB/s, while both the thinkbroadband file download test and http://testmy.net report the actual speed I usually get of around 100KB/s (or less)

- interesting is that a download will show an initial burst of speed of several MB/s, before a second or two later throttling back to 100KB/s (or thereabouts, interesting is that http://testmy.net will carry out an initial quick download in order to set the size of the main downoad, as a result the main speed test is done with a 50MB download, which I am obviously not going to wait for at 100KB/s! - this initial burst of speed is evident as the download graph that is displayed progresses)

- an ISP tech. asked me to download 2 thinkbroadband files in parallel, which I duly did, the total download speed of both files remained at the same as for 1 file, approx. 100KB/s (i.e., the download speed did not double for 2 files)

- the ISP I think it is fair to say has at this point 'deserted', a friendly second line tech. suggested I download overnight, while a call out engineer checked the modem and said everything seemed to be working fine hardware wise, it was a software problem (have not heard anything since)

To throw the cat amongst the pigeons, I have had this problem in the past, and have solved it by installing a security hardened version of UNIX (BSD) ensuring that the install remained untainted by networking only over a SSH encrypted connection. I could do the same again but it would take a good few hours (not being a systems admin. by trade and not something I have the time for at the moment).

Thanks in advance,

A quick correction, the broadband is 50Mb/s, not 50MB/s.

An additional note also:

- when downloading the speed of the download increases by orders of magnitude if a broadband speed test (reporting the full 50Mb/s) is run alongside (i.e., in parallel, returning to the slower speed when when the broadband test has ended)

There are a lot of assumptions, but would I be correct in thinking this looks very much like some sort of traffic shaping?

Edited by deleted (Sun 20-Dec-15 15:01:07)