Hannes Tschofenig writes:
> 
> On Oct 4, 2012, at 6:33 PM, Behcet Sarikaya wrote:
> 
> >> Pre-shared keys has the problem that they do not scale, and in most of
> >> the internet of things like setups you cannot really use pre-shared
> >> keys. Using raw public keys solves the scalability problem, but does
> >> not involve the certificate validation issues.
> 
> I hear this all the time and still it is wrong. 
> 
> The "magic" to scale a shared secrets-based authentication system is
> to move from a two party to a three party system.

Which is what is NOT done in the IKEv2, so that does not help. I was
not talking about the generic Pre-shared keys, I was talking about the
pre-shared keys as defined by IKEv2, which does not involve 3rd party.

> An example of a successfully deployed shared secret based
> authentication infrastructure can be found with the cellular
> communication infrastructure.

Which in IKEv2 context would not mean pre-shared keys authentication.
If you want to use that kind of shared secrets in cellular
communiations in IKEv2, you need to use EAP authentication in IKEv2.

> In context of this document this means that you have to use IKEv2
> with EAP and combine it with the AAA infrastructure.  
> Isn't it easy?

I would not classify EAP as easy, but that is besides the point, EAP
authentication is different authentication method than pre-shared keys
authentication in IKEv2.

In IKEv2 there are currently 5 (somewhat) different authentication
methods. They are in the order of easiest to implement to most
complicated to implement as follows:

  1) Pre-shared keys (i.e. shared key)
  2) Raw RSA keys
  3) Certificate based authentication methods
  4) EAP
  5) Secure password protocols

The EAP could also be before certificates, but that depends completely
what EAP method is used. For example some EAP methods do require
certificate support, which makes implementing them more complicated
than authentication based on certificates in IKEv2. Some EAP methods
are much simplier than certificates, going to be about the same as Raw
RSA keys in IKEv2.

Secure password methods could actually be also be betwen raw RSA keys
and Certificates, but that again depends which secure password method
you implement, and there is not yet enough implementation experience
for them to really specify how complex they will be to implement. 
-- 
[email protected]
_______________________________________________
Lwip mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/lwip

Reply via email to