[tor-talk] EOTK Video #3 - Deploying Onions At Scale with EOTK & OnionBalance

[grarpamp]

On Thu, Feb 23, 2017 at 12:53 PM, Alec Muffett <alec.muffett at gmail.com> wrote:
> How to set up 24 Tor + 120 NGINX daemons to mirror Wikipedia, without
> working too hard... :-)
>
> https://www.youtube.com/watch?v=HNJaMNVCb-U

A better description for this is 'proxying' for wikipedia via onion frontends.
Some years ago a number of us were actually 'mirroring' and importing
wikipedia's dumps into read-only onionland instances of mediawiki.
Most seem to have passed perhaps due to dataset growth, lack of
endangered status, moving on, etc.
Clearnet services establishing their own onion / i2p / cjdns / whatever,
is an interesting idea that's becoming more popular since the 2010's.

However there is bandwidth impact to the network compared
to the usual 'client -> exit -> service' 3-node 1-circuit exit case.
- Hosting the onion on your service interfaces costs double+
to the network over the standard exit case.
- Should never run an onion just to route that back out over
exits... that costs triple+ over the standard exit case.

If the services are also running their tor daemons as non-exit
relays with bandwidth provided back at around 3:1 of what they
handle from tor, that seems more balanced.

Adjustment is to be made of course for the preferred single-onion
services method which has recently become prominent.