IPv6 CGA and key (non-)management, was Re: How to preempt rogue RAs?

[marcelo bagnulo braun]

El 01/11/10 15:20, Gert Doering escribió:
> Hi,
>
> On Mon, Nov 01, 2010 at 02:46:55PM +0100, Shane Kerr wrote:
>> On Mon, 2010-11-01 at 07:47 +1030, Mark Smith wrote:
>>> Key management is usually more of an issue. I've wondered, but haven't
>>> looked into, whether 802.1x can be used to boot strap IPv6 SEND,
>>> facilitating a simple username/password authentication model that we're
>>> all quite comfortable with.
>> I thought the whole beauty of IPv6 CGA (horrible acronym) is that you
>> don't need key management. The address *is* the public key. (To be
>> completely correct, the rightmost 64 bits of the address is the hash of
>> the public key).
> True for "neighbor discovery" things, where you want to make sure that
> the person replying to an ND for a well-known IPv6 address is really the
> one entitled to answer (protect against ND poisoning/spoofing).  You
> need prior knowledge: you need to know who you want to talk *to*.
>
> For RAs, since you don't know who the router *is*, CGA-style protection
> ("I'm a router, and I have the key to prove that my IPv6 address really
> is what I claim the address to be") will not validate the "I'm a router!"
> bit.  For that, you need the CA stuff - someone you trust authorizes the
> router to send RAs.  Who has no CA certificate is not a trusted router.
>

that is correct as well

In SEND, address ownership is provided through CGA and authorization for 
a router to announce a prefix in a RA is provided by certificates (which 
don't need to be global, though, but you need at least a local PKI)

Regards, marcelo

> [Details might not be 100% correct, but I think the overall picture should
> be fine]
>
> Gert Doering
>          -- NetMaster