[tor-talk] Fundamental differences between Onion Routing and The Onion Router

Paul Syverson paul.syverson at nrl.navy.mil
Tue Mar 18 04:59:26 UTC 2014

On Mon, Mar 17, 2014 at 01:00:32PM +0100, Maimun Rizal wrote:
> Dear All,
> Where I can find information about the fundamental differences
> between Onion Routing (OR) and The Onion Router (TOR)?

First off Tor not "TOR" and The Onion Routing not "The Onion Router"
cf. https://www.torproject.org/docs/faq.html.en#WhyCalledTor

I respond to some of your specific statements below, but I think the
most thorough discussion of the evolution of onion routing design
including Tor (but not including any changes in the most recent few
years or so) is in "A Peel of Onion"

> what I know is The OR allow repeated use of router, for example,
> Routers used as entry relay can be used again as the exit relay when
> the path length of the network is more than two relays. However,
> this does not allow in TOR network.

If I recall correctly this was not part of any of the three main
generations on onion routing design that came from NRL (Tor being the
third). But the general idea of onion routing does not preclude
such a possibility.

> and also, TOR network has been added perfect forward secrecy,
> congestion control, directory servers, integrity checking,
> configurable exit policies, and practical design for location-hidden
> services via rendezvous point that all features were not formerly
> found in Onion Routing (OR).

Partly correct. Tor added many of these things, but configurable exit
policies (and entrance policies for that matter) were added a
generation before Tor. Also there were rendezvous servers and
location-hidden services even in the very first design, although the
Tor design changed many features of each.

> How about selection of the routers? TOR network select relays on TOR
> users side (vidalia), so the user can select multiple relays to be
> used as an entry or exit relays. the OR?

Not sure I follow what you are saying, but onion routing, including Tor,
has always had the source (client in Tor) select the nodes in the route
in an unpredictable manner. (In the very first onion routing paper
we introduced a concept we called 'loose source routing' which allowed
intermediate relays to create their own onion routes to the next hop
in the route chosen by the source. This was primarily to handle
cases when one relay could not directly reach another, although it
had other features.)


More information about the tor-talk mailing list