View Single Post

Default 

December 1st, 2011, 02:15
The math of torrents is this: total download speed = total upload speed.

So lets see how that plays out when we plug in Eisberg's assumptions. Specifically he assumes the average downloader (peer) gets 14GB in 6 hrs. That in turn equals 2.333 GB/hr which calculates to be 0.03888 GB/min, which calculates to 0.00064805 GB/s. IF we assume a seed/peer ratio on average of 1/1, i.e., one to one, (bear in mind that is a very optimistic assumption as most torrents have fewer seeds than peers), then we find that the average seed (uploader) in these torrents must be uploading at an average speed of 5,567 kilobits/sec (see conversion table at http://www.speedguide.net/conversion.php).

Granted I live in the US which has rather slow internet speeds; but I do have a high speed internet connection; and my MAX upload speed is 384 kilobits/sec. But even with this I cannot actually upload at 384 kb/s because part of what I am uploading is headers (each internet data packet includes a header that specifies where the data is going and other stuff); so my max upload speed is actually about 350 kb/s (assuming a 10% loss due to headers). And at that upload speed I have used up my entire bandwith — I cannot do anything else on the internet — and this also assumes that my torrent client is somehow 100% efficient in being able to use all of my available bandwidth (a virtual impossibility).

Net of these calculations; the average seed in Eisberg's hypothetical torrent transfer is continuously and constantly uploading at a speed of 5,567 kb/sec for 6 hours. That is an upload speed of nearly 16 times my max upload speed.

These assumptions are totally unreasonable. The total downloads need to be cut by at least a factor of 5 to 10 (probably more).
Last edited by RPGFool; December 1st, 2011 at 02:31.
RPGFool is offline

RPGFool

RPGFool's Avatar
Watchdog

#14

Join Date: Aug 2011
Posts: 191