[tor-dev] Fwd: Re: Support for mix integration research

[Sebastian G. <bastik.tor>]

I'm forwarding this to the list, because it might contain relevant
information. Mostly within George's reply.

Initial reason to not send my reply to the list was to not bother too
many people with seemingly irrelevant questions on my part.

I agree with the reply given in the forwarded message. It has to be
considered what 'it' does to the network.

Best Regards,
Sebastian G. (bastik)


-------------- Forwarded Message --------------
Subject: Re: [tor-dev] Support for mix integration research
Date: Thu, 25 Feb 2016 19:58:56 +0100
From: George Kadianakis <desnacked at riseup.net>
To: Sebastian G. <bastik.tor> <bastik.tor at googlemail.com>

"Sebastian G. <bastik.tor>" <bastik.tor at googlemail.com> writes:

> 25.02.2016, 13:08 George Kadianakis:
>> b) Does this increase the hard disk or memory requirement of people running Tor
>>    relays? That is in high-latency mode, will Tor relays need to store client's
>>    traffic for longer?  How does that impact the memory and hard disk
>>    requirement of people running Tor relays?
>
> Hello,
>
> this has been intentionally sent off list.
>
> I just don't understand the question. How would you delay traffic
> without storing it for longer? Whenever in memory or disk, the data that
> are being delayed have to be stored somewhere, as far as I understand.
>
> Could one delay traffic on a wire? E.g. like with a truck that drives on
> a road between two storage facilities where you couldn't say it is
> already stored in the facilities.
>
> I fail to see how it would be possible to have a system that delays
> something without holding back whatever is delayed.
>

I agree.

I'm wondering how this increases the memory and hard disk requirement of
running
a relay. And how much would the current relays contribute to such a
high-latency
system? And how much memory / CPU will be required to contribute
meaningfully to
this. And what about DDoS opportunities?

All these things will need to be thought of so that we don't overwhelm the
current Tor network by introducing this high-latency mode. Because if
that's the
case, we should probably have a separate network for the high-latency mode.

Pity you replied off list, this conversation is of sufficient interest.

Cheers.

> If you currently take in x and give out y at the same rate, what needs
> to be stored is what has to be buffered. If now the rate of y decreases
> while the rate for x remains the same, more stuff needs be buffered.
> Unless you reduce the rate for x, more storage is required.
>
> Sorry, I just don't get it. Please show me where my error is. Can be
> short. :)
>