On Mon, 13 Apr 2015, Stephen Farrell wrote:
I'm not sure if your point was meant to relate only to DHCP
setting the DNS server IP, but if not then I have a question...
Nope.
On 13/04/15 21:21, Paul Wouters wrote:
If you have an attacker on the last mile, there is nothing you can do.
Passive only protection against the last mile is a wasted effort. On
the last mile, there are only active attackers.
Do you have evidence of the absence of passive attackers and if
so with what definition of last mile?
Googling for "wifi mitm software" comes up with dozens of scripts that
are trivially run. The barrier to entrance is zero.
Noting that an active attack is possible does not IMO mean that
defence against a passive attack is worthless.
Agreed, provided there is an actual difficulty difference between the
two. For wifi at starbucks, there is really no reason for an attacker
to stop at being a passive attacker.
that would need to be made IMO. And the counter argument is that
the probability of an active attack may differ significantly from
the probability of a passive attack.
That is surely true for passive attackers upstream and further in the
core. but I do not think that applies to the last mile. The only reason
for not doing an active attack is because it is not needed, not because
there is any technical difficulty in performing an attack using a
downloaded program with zero computer knowledge.
Basically, I think you're overstating things in the quoted text.
That is quite possible, I have no numbers.
I'd like the WG to focus on building an encryption method to a remote
DNS server that can be authenticated regardless of the last mile. Then
adopt a "better than nothing variant" of that for the last mile. And not
build something from scratch just for the last mile. I strongly believe
that's not worth it.
Paul
_______________________________________________
dns-privacy mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/dns-privacy
