[cabfpub] Pre-Ballot 164 - Certificate Serial Number Entropy

Doug Beattie doug.beattie at globalsign.com
Thu Apr 28 13:45:24 MST 2016


I also think 6-months is also a good guideline unless it’s a security threat.  Is the current serial number requirement suddenly a security issue that needs to be resolved in the next month?

Doug

From: public-bounces at cabforum.org [mailto:public-bounces at cabforum.org] On Behalf Of Ryan Sleevi
Sent: Thursday, April 28, 2016 4:23 PM
To: Rich Smith <richard.smith at comodo.com>
Cc: CABFPub <public at cabforum.org>
Subject: Re: [cabfpub] Pre-Ballot 164 - Certificate Serial Number Entropy



On Thu, Apr 28, 2016 at 1:15 PM, Rich Smith <richard.smith at comodo.com<mailto:richard.smith at comodo.com>> wrote:
I do think this brings up a good point though.  This has come up before under other ballots requiring code changes to CA core systems.  I think that any change requiring such code changes should have a minimum lead time of 6 months from passage of the ballot before becoming mandatory, unless it is deemed to be a security threat sufficient to require more immediate action.  Admittedly I do not have the technical expertise to know if this is such a case.

Could you explain how you chose six months, as opposed to one month (the time for legal review for Final Guidelines) or, say, three months?

Our experience with responding to security threats, at Google, in Chrome, and in conjunction with other vendors (such as through efforts like Project Zero), is that, in the worst case, the ability to respond to a security threat in a timely manner is directly related to the ability to release and deploy new versions and products of code. That is, even if we were to say that an incident is security related, and thus perhaps requires 14 days to effect change, there will be some portion of CAs who, through the structure of business operations and third party engagements, will have difficulty responding to anything outside of their contracted timeframe - such as the six months you propose. By setting the expectation lower, such that it's clear what the 'worst case' scenario may be, the overall security of the ecosystem improves.

That said, I would also argue it's difficult to quantify what represents a change to CA core systems, because that varies largely depending upon how the CA has structured their operations. Because that's such a subjective measure, and one for which there is an unfortunate long-tail of CAs who are unable to take any necessary precautions in a timely fashion, setting that as the bar may be disproportionately disadvantageous to security.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: https://cabforum.org/pipermail/public/attachments/20160428/51f10d10/attachment-0001.html 


More information about the Public mailing list