Re: [lisp] Comments on draft-ietf-lisp-eid-anonymity-00

[Dino Farinacci <[e-mail]>]

> Hi,

Thanks for the comments Rene!

> 'draft-ietf-lisp-eid-anonymity-00' does not mention any authentication of the LISP-nodes/xTRs with the map-servers.
> That way a Man-In-The-Middle-Attack can be run by sending bogus map-requests with the MITM-attacker's RLOCs.

Well we do have the regular shared-key authentication documented in draft-ietf-lisp-rfc6831bis. But we can also put a reference in for draft-farinacci-lisp-ecdsa-auth if you are looking for stronger ways to authenticate.

> I suggest to allow only CGA-addresses.

There is a reference in this section:




> Benefits:
> 
> 1. Privacy:         Dynamic eEIDs by re-keying
> 2. Reachability: Static      eEIDs by static key-pairs allow direct connections without third-party services relaying (e.g. Facebook ;-) )

Yes, understand.

> 3. Security:
> 
>   1. Signed map-requests
>   2. eEID-based authentication/authorization with static keypairs
>   facilitates administration
>       (e.g. road-warrior authentication/authorization can be done with
>   a firewall by eEID filtering
>       without any upper-layer authentication/authorization like
>   usernames/passwords involved)
> 
> Privacy/security considerations:
> 
>   1. Don't use dynamic and static eEIDs on the same RLOC!
>   2. Always use payload encryption to avoid deanonymisation by
>   Deep-Packet-Inspection

Right, this is all in draft-farinacci-lisp-ecdsa-auth. Have you looked at it?

Dino