Re: DNSSEC architecture vs reality

[Masataka Ohta <[e-mail]>]

Patrik Faltstrom wrote:

> Because doing signing is viewed as a cost, and there must be a benefit in doing the signing.

True.

As PKIs, including TLS and DNSSEC, rely on CAs as untrustworthy
TTPs, they are, as was demonstrated by diginotar and google,
subject to MitM attacks on CAs.

That is, PKIs are only as secure as plain Internet with ISPs
as untrustworthy TTPs.

As such, there is no point to waste the signing cost.

 > This is why validation must come first. The cost is low,

But, there is no true security for the cost.

						Masataka Ohta

PS

It may still be a good idea to use Diffie Hellman key exchange,
because it is secure against passive attacks.