Seems very strange to me, but I've run a few tests, and would love some ideas of other tests. I have a linux box here running Deluge, where I most of my bittorrent activity happens, and a Windows box where I do less of it. I've performed these tests:
Downloading the same torrent, over the same forwarded ports on windows and linux.
Downloading the same torrent, over the different forwarded ports on windows and linux.
Downloading the same torrent, using the hardware from the windows box on linux (using a livecd).
Downloading the same torrent, Using the linux box, with a windows VM running in virtualbox. (I know, not a REAL test)
The results are always the same. Under windows, using uTorrent 3.1, I see the traffic spike and then slow to a trickle for the remainder of the download. Under linux, using deluge 1.3.3, the traffic maintains it's high speed throughout the download. I'm baffled by this behavior. Have I overlooked some obvious test that I should be doing? How would it even be possible for my ISP to profile windows-based bittorrent activity and not linux traffic?