Thank you very much for the pointer to that thread, super interesting! On Mon, Oct 13, 2014 at 6:32 AM, isis <i...@torproject.org> wrote:
> The problem is this: All clients fetch information about all the relays in > the > network from the Directory Authorities/Mirrors, and these fetches take up a > certain amount of bandwidth. If the relay is too slow, the bandwidth > provided > by that relay does not compensate for the directory fetching bandwidth > used to > tell people about the relay, and thus it is actively harming the network. > This is probably a silly question, but why don't nodes themselves gossip in a P2P fashion about other nodes they've seen or heard about, in order to avoid taxing directory authorities and mirrors? I'm sure that there are security implications for this, but perhaps they could be overcome by preferring nodes in different IP blocks and with diverse traceroutes, etc. > Additionally, since Tor processes are normally CPU-bound, most relays > aren't > able to use all their available bandwidth with a single Tor process. > Running a > relay on ARM (or likely any other mobile/low power) CPU will only further > limit how much traffic your relay is actually pushing. > Is this because of the outer encryption wrapper layer decryption? > Additionally, if you're attempting to do this with Orbot on an Android > device, > you'll run into issues with Android's process management system and the Tor > process randomly dying unexpectedly. This means that you are providing an > unreliable, flapping relay which is actively messing up other people's > connections through the Tor network. > Is this an issue on all devices, or just some of them? Perhaps oorbot could refuse to run, or at least display a warning if it looks like the device is flaky. -- tor-talk mailing list - tor-talk@lists.torproject.org To unsubscribe or change other settings go to https://lists.torproject.org/cgi-bin/mailman/listinfo/tor-talk
