On Friday 01 Nov 2013 14:39:28 Gordon Morehouse wrote:
Completely aside from the ethical and censorship-related buzzsaw you're about to run into for posting this (perennial) question, I believe some actual developers on Tor have written a paper about the problems with Bittorrent et al (and I think there's a more specific one than the Why Tor Is Slow[1] paper) but I can't currently find it. Anybody know?
https://svn.torproject.org/svn/projects/roadmaps/2009-03-11-performance.pdf
NB: the above paper is from 2009.
I've just had a quick scan of that paper and it makes for an interesting read. :) I'm going to go away and read it properly but a couple observations.
2.3 Throttle certain protocols at the client side I agree this is not a good idea BUT it sparked off another idea in my head. In order to get better utilisation of slower relays, would it be worth introducing a behaviour whereby "slow" circuits are deliberately built for low volume traffic?
For example, sending email and IM messages doesn't (usually) require a huge amount of bandwidth, so when the Tor client detects that a user wants to send/receive data on certain slow ports such as POP3, IMAP4, MSA and Jabber it deliberately builds a "slow" circuit to handle that traffic. Obviously it would have to be port based, but since people tend to send data on well-known ports, it shouldn't be an issue.
I think this would play well with the circuit-bonding work here
http://freehaven.net/anonbib/#pets13-splitting
3.1.2 Better Support for relay operators This caught my eye: "We lose relays when the operator reboots and forgets to set up the relay to start on boot." Does installing the .deb package (for example) not configure Tor to start on boot, in the same way that Apache would be? I ask because I haven't rebooted my VPS yet. :p
3.1.3 Facebook app to show off your relay I liked this bit: "Opportunities for expansion include allowing relay operators to form “teams”, and for these teams to be ranked on the contribution to the network. (Real world examples here include the SETI screensaver and the MD5 hash crack challenges.)"
What would be really interesting would be to find sponsors (read: hosters) willing to put their name to it and gain/risk some publicity.
3.2 Funding Relays Directly 100 10Mb/s relays at $10/month is $12K a year, not $120k/year. :)
Best,