At our institute, we will soon need to routinely share large volumes of data (multi-terabyte range).
-
Would it make sense to use BitTorrent for this task?
-
How large of a CPU/memory overhead is to be expected, compared to common FTP servers?
-
Is it possible to achieve speeds matching a direct FTP transfer when copying from exactly one BitTorrent peer (the original storage server) to another?
Thank you very much.
Best Answer
To mitigate issues regarding points 1 and 2, if you can split the data into smaller chunks and have separate torrents for each chunk you might find the size of the data easier to handle.
Also note that you will need to regenerate the torrent metafiles if any data in the file(s) they cover is updated. If small parts of the data change without the rest changing, you probably find rsync to be a much more efficient solution.
How large are the files in the dataset and what is the spread like (several multi-gig files?, many smaller ones?, ...)?