ietf-mailsig
[Top] [All Lists]

Re: DKIM: Key revocation

2005-07-15 11:15:05


On Jul 14, 2005, at 11:29 PM, <domainkeys-feedbackbase02(_at_)yahoo(_dot_)com> wrote:
--- Earl Hood <earl(_at_)earlhood(_dot_)com> wrote:

There should be some way to indicate that a key has been revoked.
It seems right now, revocation is indirect, by removing the appropriate
DNS entry.  However, a verifier does not know if the key was revoked,
temporarily unavailable, or never existed.

Hmm. I'll have to swing back on the draft, but at one stage the plan was to use the DK revocation method of continuing to advertise the Key RR with an empty
public key value.

Would that satisfy your concerns regarding explicit revocation?

Revocation could be seen related to the replay issue. When there is a large domain sharing the same key among many users, such key revocation creates problems. Having a key per-user creates different and perhaps equally serious problems. Revocation by modifying or removing the key also requires short TTLs for the DNS record, which exacerbate the effect of using a large number of individual keys.

In the mass-reputation draft, I suggested the optional use of a u=<revocation-identifier> in the signature header to provide two benefits. One benefit would be for the reputation service, where there would be a means to identify abusive sources on the large domain that consolidate any number of abusive messages, and to see these accounts being handled by the providers. The other benefit prevents there being a conflict of interest with respect the same independent entity providing finger-prints on replay messages, and domain reputations. The revocation-identifier would enable each provider to establish their own individual account revocations, without relying upon third-party threat information.

It is already common for providers to identify accounts using some opaque manner. Adding the u=revocation-identifier offers a means to standardize this technique. This would enable a means to eradicate one of the most troubling problems plaguing email. There are methods that can both inhibit replay attacks and lessen the overhead needed to address the replay issue as well as protecting the network resources.

-Doug






<Prev in Thread] Current Thread [Next in Thread>