Scott Fluhrer (sfluhrer) writes:
> If the requirement for AES-256 is to handle the scenario "someone
> gets a quantum computer", then in that scenario, there is no
> realistic DH group size that is secure.

That we do not know until we know what those quantum computers can
really do... I have not seen anybody saying how many qbits you need to
break MODP-2048. Most of the things I have seen talks about factoring
RSA, and even then they do not provide numbers.

draft-hoffman-c2pq also says that we might have machines breaking
AES-128 before than we have machines that can break Diffie-Hellman,
i.e., it is most likely easier to make machine running Grover's
algorithm than machine running Shor's algorithm. 

> Hence, I personally see no point in allocating IANA numbers for the
> larger than 8k MODP groups. The only scenario I can think of where
> they might be useful would be one where all of the following apply:
> 
>       - We believe that there's an adversary that can perform
>         significantly more than circa 2**128 computations' 
>       - We are not concerned with adversaries with a Quantum Computer
>       - For some reason, we don't want to use ECDH.

Or cases where customer requires your product to include support
256-bits of security level. Perhaps they read from the paper that 128
bits is too little if someone gets quantum computer and thats why
256-bit security is what you need. I.e., there is no real security
reason for that, just to be able to tick check box saying "support
security level of 256-bits".

To provide that you need AES-256, and SHA2-512 for symmetric parts,
and then either P-521 or MODP-15360 for Diffie-Hellman. If you do not
have ECDH in your implementation it is much easier to add just
MODP-16384 in configuration especially as the main reason is to tick
that check box.
-- 
[email protected]

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

Reply via email to