/ Review


Method:

You know nothing about torrents until you try to exploit them, in order to do something it was not intended for. You would think that setting up torrent clients to count the number of seed and peers would be an easy thing. Not at all, even though torrents are designed to be downloaded and shared. They do not like it if you are constantly downloading at very low speeds over extended periods. In fact some torrent sites would ban people who used 56kb modems, exactly for this reason, back in the day. (I know, what is a modem?)

In the beginning everything was great! The torrents were activated when I needed to take results and then switched off for the rest of the day. Of course, because we are an anti-piracy group, all the torrent data was deleted afterwards. I knew about torrent trackers reducing your download speeds according to your share ratio. But as I was only turning the torrent on for about half a minute, per torrent, and I only needed the swarm count, I assumed that it wouldn't affect me, how wrong I was.

After about a month I really began to notice the problems with torrents. They began acting sluggish and generally not connecting to swarms in trackers. While I admit this means that it doesn't give a full picture, at least the result is undervalued and not overvalued.

As you can imagine, even though I am not downloading the material, I am still taking up some bandwidth. Over time trackers, labelled me as a leecher, and connections became closed to my bit torrent client. This was why I needed a Ratio Faker, uSerenity, these programs send out fake information, telling the trackers that I am actually uploading, sharing, instead of downloading.

This is frowned upon in the torrent community; imagine it, moral standards and illegal downloader sounds like an oxymoron! However, the ratio faker only has to upload really small amounts to offset the download. After a week of tweaking, I have it set just right, after all I am not out to mess around with torrents, just observing.

But still more problems occurred, as my ISP, began actively disrupting with torrent traffic. But it wasn't a problem and was solved quickly, still it cost me a few days of worry. When torrents don't work right, and you are monitoring it, it has a tell, and is very easy to spot.

At the same time, I was having trouble with the sample size, as you can imagine 60+ (At that time) torrents at half a minute each means half an hour or so. On top of that, the sample sizes where only growing, at about 10 games a month, I soon realised that I would be spending too long just collecting data. (I looked seriously for some kind of automation program, but nothing exists.)

To offset this problem, until I get some automatic login program for torrent results, I will have to make some torrents redundant. Shame, because I would like to follow torrents to the end, as I suspect that the real popular ones carry on for years.

However the problem still remained that the sample size would still increase drawing more bandwidth. The results were fluctuating wildly due to the stopping and starting of torrents, so I had to come up with a whole new method of using the bit torrent client. After trying many different methods, the one that I found that works the best is:

We set all the torrents to download, but set the global download rate of 1kb/s. 62 torrents are all trying to share the tiny bandwidth and there isn't enough to download anything. However this still means that the torrent connects to the tracker and still receives updates about the P2P network and swarm.

When I want to take results, I allow them to run at full speed, just to receive any last minute update about the swarm. But when the test is over all the torrents go back to a drip fed situation. The torrents are run in small batches for the testing phase, and of course all data downloaded in that time is deleted at the end of the test cycle. , I have found this to be very effective and it seems to be running without any problems!

General Information about Torrent Watch.
  • We are using uSerenity 2.0.4!
  • The graph display was coded by amCharts, which you can visit here: http://amcharts.com/
  • Torrent results are taken at set times each day (10am, 2pm, 8pm, 1am). Results are taken within +/- 30 minutes of that time.
  • ISP sometimes have a nasty habit of interfereing with torrent data transmission!
  • Some games have more than one torrent, but normally there is only one with considerably higher Seeds and Peers.
  • 27/08/2010 01:00:00 AM: Results missed. Due to the Internet Service being down, was unable to Remote Desktop into Torrent Computer!
  • We only measure the seed/peer information.