[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Certificate suspension




Thanks for the insight.

Personally, I believe it is the "temporary" that is the cause of discomfort. As you already know, the CA is effectively saying "We can not guarantee proof of possession of the private key to this end entity", and then it is removed without applications ever knowing. This is of course, only by personal experience.

Rather, I would be more comfortable if something like invalidityDate were added, only in the form of a date range, so the CA then says "We can not guarantee proof of possession of the private key to this end entity between the dates of [beginHold] and [endHold]. Further, eliminate the ability for the CA to remove the entry from the CRL, but allow for replacement in the event it is actually revoked in the future.

I don't feel the CA should not be in the business of suspending a certificate to deny access to applications. I feel that that is more the application/authorization responsibility. (i.e., certificate vs. key suspension)

Perhaps this seems a bit like a Rube Goldberg process, but in the end, it gives a true technical capability to accept, or outright deny as revoked depending on the application/business needs. That's only my opinion though. And of course, everybody has one ;)


Denis Pinkas wrote:
Russ and Todd,

I do not share your opinions.

The U.S. Treasury does not support suspension mainly due to obvious issues encountered with digital signature related transactions, and future validation of those transactions. The problem is, if you are willing to honor signatures produced using certificates from an infrastructure that does support suspension, you wind up in the same boat. It has been a topic of much debate.

At the very begining, I was reluctant to support suspension, but there is no security flaw both for authentication and for non repudiation.

In general, if a relying party application does not support suspension, then it will treat suspension as (definitively) revoked.

In the case of non repudiation (which mandates the use of either a time-stamping or a time-marking mechanism) there is however a slight difference:

- If the relying party application supports suspension, the (electronic) signature will be considered as (temporary) invalid. In some cases, it MAY try again *at a later time* to gather new revocation information which demonstrate that the electronic signature is *now* valid. These applications may thus know when it is no more necessary to attempt to validate temporary invalid signatures. - If the relying party application does not support suspension, the (electronic) signature will usually be considered as (definitiveley) invalid. However, this class of applications could be designed to support suspension, without supporting the suspension extension. The relying party application MAY attempt to validate at a later time, e.g. two or three days later, electronic signatures that were invalid because the signer's certificate was revoked. So they may end up with the same result, but not necessarilly within the same time frame.

Denis

--


Regards,

Todd E. Johnson