[tor-talk] Layer-7 DoS Attack Against WWW Tor Hidden Service

[George Kadianakis]

bob1983 <bob1983 at protonmail.com> writes:

>>> Is there a way to limit resource usage originated from a single Tor circuit?
>
>> There is no such functionality right now I'm afraid. People have been
>> wanting some sort of functionality like that for a while:
>> https://www.hackerfactor.com/blog/index.php?/archives/777-Stopping-Tor-Attacks.html
>> but we haven't had time to develop/design something.
>
> The first possible solution quickly came to my mind, was to do what i2p was
> doing for years: assigning a placeholder IP address based on the 32-bit hash of
> the circuit ID. It is not an ideal countermeasure, but it does work for simple
> rate-limiting purpose and compatible with existing IP-based applications and
> servers.
>
> I've just checked the source code.
>
> https://github.com/i2p/i2p.i2p/blob/920b14212fa80a3a0e92d6e919fdae7e39ed22d5/apps/i2ptunnel/java/src/net/i2p/i2ptunnel/I2PTunnelServer.java#L739
>
> and it turned out that this feature from i2p was actually based on this patch
> purposed in tor-dev mailing list!
>
> [tor-dev] Patch: Hidden service: use inbound bind-address based on circuit ID
> https://lists.torproject.org/pipermail/tor-dev/2014-March/006610.html
>
> I don't know if it's still worth to try with this approach, or developing a
> separate API should be the right way to go. Any other insights, anyone?
>

Thanks for this information bob1983. I opened ticket #24298 to handle
the generic issue of DoS attacks, and also opened #24299 to investigate
the I2P feature you mentioned. Hopefully we can find some time to work
on this, or it might give the community a place to design stuff.

I'm also wondering how the I2P community is using that feature. I have
asked some I2P friends and waiting for answers.