> Wireguard does not have a control-plane, this means that Wireguard nodes need 
> to be manually configured before being able to exchange packets. Manual 
> configuration typically involved provisioning public keys using out-of-band 
> mechanisms. In this context, we have architected and prototyped a 
> control-plane for Wireguard using LISP, this enables automatic and secure 
> retrieval of public keys using LISP.

Sounds good Albert. I have looked at Wireguard in the past and agree its great 
stuff.

Note the LISP-decent stuff allows the wireguard nodes to be their own mapping 
system. So you can continue to use and deploy Wireguard in a decentralized 
manner.

Also note, you can distribute public-keys using the draft-ietf-lisp-ecdsa-auth 
(and draft-farinacci-lisp-decent). Colin and I are working on distributing 
public-keys by the nodes that generate their own key-pairs without a need for a 
third-party trust anchor.

> This raises -hopefully- interesting questions, how should LISP support 
> multiple data-planes? In this context Wireguard can be seen just as another 
> data-plane. Additionally, Wiregard provides a secure data-plane, can we learn 
> something from them? 

Use the LCAF Encap-Format Type, so an ETR, when it sends a Map-Reply (or the 
mapping system) to indicate which data-planes an ITR can use to encap traffic 
to the ETR.

Note that if Wireguard wants to rekey the data-plane keys, it can use 
RLOC-probing DH key exchange documented in RFC 8061.

Let me know if you need any help or clarification.

Dino



_______________________________________________
lisp mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/lisp

Reply via email to