BitTorrentPerformance

From TheoryOrg
Jump to navigation Jump to search

e have been commenting about relative performance between versions 3.1 and 3.0.2 a lot. This is a very complicated issue, so I'll explain in detail. At the bottom I'll explain what I'm going to do about the new problems, they're all quite fixable. --Bram

See message 1260.

In general though, when people complain about performance it has nothing to do with BitTorrent itself. It's much more likely that the transfer is just slow for natural reasons, there is not enough bandwidth.


BT is slow as anything now, the seed trys to allocate parts to everyone, meaning on the last 5 swarms I've seen with between 20 and 500 people. We all slowly crawled in tandem to 88.1%,88.2% then we all became seeds together to serve the remaining 2 or 4 people on 99.99%. Then the swarm collapses again. Making seeds pointless now.

  • This is typically only the case of lots (hundreds) of people jump on a torrent before there is more than one seed. Basically everybody is at the mercy of the one seed, and their upstream cap.

You see the people who get the files in a 1/2 hr while you wait 8 hrs? Make sure your firewalled is configured to forward your ports and make sure you're uploading as much as you can.


I am interested in setting up my own tracker, and hopefully, it will attract many users. What kind of performance do the trackers usually require? What kind of bandwith overhead is there just for being a tracker? How about ram and cpu? Do these things scale linearly?


Part of [[]]


Last edit: Fri, 14 Jul 2006 13:48:11 -0700
(WikiWord)
Revisions: 12