Garlic Routing

The Basics

The term Garlic Routing has a diverse history of varying interpretations. As it currently stands, Monero defines Garlic Routing as the method in which Kovri and I2P create a Message-based anonymous overlay network of internet peers.

The Garlic Encryption of Garlic Routing is similar to the Layered Encryption of Onion Routing and effectively conceals the IP address of the sender and secures information sent from the sender to its Destination (and vice-versa).


In written form, the term Garlic Routing can be seen as early as June of 2000 in Roger Dingledine's Free Haven Master's thesis (Section 8.1.1) as derived from the term Onion Routing.

As recent as October of 2016, #tor-dev has offered insight into the creation of the term Garlic Routing:

Nick Mathewson:

[I think that there was some attempt to come up with a plant whose structure resembled the 'leaky-pipe' topology of tor, but I don't believe we ever settled on one.]

Roger Dingledine:

during the free haven brainstorming, there was a moment where we described a routing mechanism, and somebody said "garlic routing!", and everybody laughed. so we for sure thought we had invented the name, at the time.

Note: permission to use the aforementioned quotes was granted by Nick Mathewson and Roger Dingledine

In-depth Information

In technical terms, for Kovri and I2P, Garlic Routing translates to any/all of the following:

Note: though Tor uses Layered Encryption, Tor does not use ElGamal and is not message-based.

Read more in Garlic Encryption.


  • In terms of Onion/Garlic Routing, another way to envision layered encryption is by replacing the onion/garlic with a Matryoshka doll - with each outer/inner doll having a lock and public key to the next/previous doll
  • For more technical details on Garlic Routing, read the Java I2P entry on Garlic Routing