[NTLUG:Discuss] BitTorrent on Libranet

[Ralph Green, Jr]

Howdy,
  Until a change in the security policy about a week ago, I had been
using BitTorrent a fair amount for several weeks.  I'll answer your
question and add a couple of other comments.

 When you ask about percentage hit, I see two ways of answering this. 
One is the bandwidth consumed.  I would estimated that BitTorrent
doubled the bandwidth needed when there were only one or two people
uploading a file.  The second method of judging is wall clock time. 
Here the impact is much greater.  It seemed to take much longer, maybe 4
or 5 times as long.  I can remember a couple of 300 meg files that took
2 days each to get.  If I had been able to ftp them, I would have.

  I don't want to sound negative,  I think it is a very promising
distribution technique.  When there were a modest number(30 to 100) of
uploaders, the bit just seemed to fly in.

  I did one download where there were a large number of participants. 
It was a home made computer show that got listed on slashdot.  I had
read that BitTorrent did not scale well above 2000 participants for a
given file.  My experience confirmed that, albeit with a sample size of
one download.  The download crawled until late at night when the number
of participants dropped.  And I had to stop for a while because it
disrupted the office.  Even though the download rate was small, there
was so much protocol traffic that the T1 in the office was bogged down
and people complained.

  I also experimented with various settings for my maximum upload and
download rates.  Setting a larger upload rate really made a difference
for me in getting faster downloads on a moderately active torrent.

 I only downloaded stuff that was legal to distribute, mostly Linux
ISOs.  If you use it for anything else, keep in mind that it is not
anonymous and you can be tracked.

  I look forward to getting broadband at home someday and I'll
experiment more with legit P2P.
Have a good day,
Ralph

On Wed, 2004-03-31 at 13:46, Robert Citek wrote:

> Any guess as to what the "efficiency hit" would be?  That is, how much 
> slower would BT be vs FTP or HTTP assuming only a single user and all