Can bittorrent peers handle a large number of unoccupied torrents

I am considering using bittorrent for a big data distribution problem where the data source is petascale and users will want up to a few terabytes. Some details

  • The number of torrents, potentially in millions
  • torrent sizes from 100 MB to 100 GB
  • A stable set of clusters around the world that can act as seeders, each of which has a large subset of common torrents (say, 60% on average)
  • A relatively small number of concurrent users (less than 100) who want to load on average several terabytes of data.

I expect that the number of active torrents will be small compared to the total available, but the quality of service is important, so for each torrent or some mechanism for launching new seeders, there should be several seeders.

My question is: can bittorrent clients handle a huge number of torrents, most of which are idle? Do I need to break down the flows by seeders in the cluster or can each node sow all the torrents to which it has access? Which client will do the best job? Are there any tools for managing seeder clusters?

I assume that trackers can be scaled to this level.

+5
source share
4 answers

There are two main problems:

  • () , .
  • bittorrent ,

, , 1 , 30 , 1 . , . 1 GET , , 400 100 ( , ), 111 / 28 / . , , TCP , 40 40 .

, UDP. -, . 100 , , , 60 . 28 / 16 /, . udp (, , ).

, , , .

, HTTP- . bittorrent HTTP-, (URL- .torrent).

, , , . .

libtorrent rasterbar, , .

, .

+4

, Hekate , , - , , .

+3

- , ( ), , . ().torrent , . , - , .

rTorrent ( , , .torrent), xmlrpc, . , .torrent , 24 , .

+1

, , . .

, , , , , . , TCP- , .

BitTorrent:

, . - , , , , .

Bittorrent v1.0:

, . , ( info_hash).

, , . , , , , . . , .

0
source

All Articles