Re: [TLS] OCSP must staple

["Salz, Rich" <[e-mail]>]

> Well, if the server *is* OCSP stapling capable, there is little harm in having a "must staple" certificate.

Except that they are depending on the CA's OCSP responder, and trusting (worrying is a more accurate term) that the browsers will do the right thing, whatever that is.  I'll be pretty surprised if the "select few," as you put it, will want must-staple certificates. 

> but larger problem lies upstream, working revocation for the masses.

Which masses?  For what? We're going through an exercise of checking the SSL certs on the true origins of many of our customers.  You'd be sadly surprised how many of them are expired, have wrong SAN fields, unknown CA's, and so on.  And even sadder if you knew how many said "we'll deal with it later, if at all."  And yet, the web commerce still marches on.

I  think the general public is put more at risk by erroneously-issued certificates, whether by accident or malicious or "national security."  Luckily, it appears that Certificate Transparency seems likely to narrow that window of exposure down to a fairly short time.

	/r$

--  
Principal Security Engineer
Akamai Technologies, Cambridge, MA
IM: [e-mail]; Twitter: RichSalz