Re: [Pidloc] PIdLoc Webex

[Tom Herbert <[e-mail]>]

On Tue, Dec 4, 2018 at 8:53 AM Dino Farinacci <[e-mail]> wrote:
>
> > Questioned explicitly by Dirk Luigi pointed out that LISP so far has not dealt with privacy problem but secured content by encryption and protected mapping request from attacks
>
> This is a broad statement. Can someone be more specific.
>
> If DTLS is used for LISP protocol messages and LISP-crypto is used for encapsulation than both the LISP control-plane and data-plane support privacy. Note that if you don’t use DTLS, one can send control-plane messages over LISP-crypto. So the architecture can support privacy.

DTLS could be applied to any UDP tunneling protocol, but that's not
really where the privacy problem lies. Presuming that as much of the
packet is encrypted as possible, then the question becomes what can be
inferred from the non-encrypted portions of the packet to breach
someone's privacy. The obvious data that could be exploited are the IP
addresses in a packet. There's at least two possible exploits: 1)
Addresses allow correlations between two different flows that they are
originated by the same user 2) Geo-location of a user can be deduced
from observed addresses. The actual identity of a user isn't
immediately available in addresses, but there's a fairly simple method
to deduce identity using #1.

Tom

>
> Dino
>