James G. Sack (jim) wrote:
This /. article touches a discussion that (IIRC) occurred here about
whether using Amazon for bittorrent seeds would actually work -- or
would all the traffic naturally migrate to Amazon as the biggest pipe?
http://slashdot.org/articles/08/03/08/042227.shtml
A Norwegian broadcasting company NRK, did this experiment and figures
torrent cut their costs to 1700/41000 (~4%) of their estimated
non-torrent costs!
Anybody see anything to question about this calculation?
I can't read the Norwegian, but it doesn't seem they don't have any
details about their AWS setup, merely mentioning S3.
Two things:
1) Amazon generally has some of the best bandwidth prices around bar
none. Nobody I have been able to extract a quote out of comes close.
For something like BitTorrent which is basically just vacuuming up dumb
bandwidth, this is pretty close to ideal.
2) Amazon would become the main pipe given infinite bandwidth and zero
latency. In the real world, however, Amazon presents neither. I'm sure
that Amazon has limits on both.
If nothing else, the compute power of an EC3 instance probably limits this.
So, if too many people were using you as a direct download, all you need
to do is reduce the number of EC3 instances running BitTorrent. Latency
goes up, bandwidth goes down, and people start pulling the other seeds.
-a
--
[email protected]
http://www.kernel-panic.org/cgi-bin/mailman/listinfo/kplug-list
