I was doing quite some PKI projects in my former life. One of the key themes during the policy discussion and then afterwards in the implementation was always the way somebody can revoke a certificate and then how the revocation was communicated. Shall OCSP be used or shall we stay with the good old CRLs? Well, often it did not make any difference as a lot of application did not care for revocation anyway and just checked the certificate trust chain. In the early days we did some tests with browsers and they tended not to look at CRLs by default as they were too big to download and it would delay the handshake â€“ and how may certificates will be revoked anyway?
That changed over time as the speed on the lines grew and downloading a file was not too big a deal anymore or OCSP could be used.
And then Heartbleed came â€“ and with Heartbleed the need to revoke certificates on a massive scale and with that huge CRLs and applications which should check the revocation state of the presented certificate.
I have no real insights into how this will evolve but it will definitely be worth keeping an eye on.
- Chrome users oblivious to Heartbleed revocation tsunami (news.netcraft.com)
- The Hidden Costs of Heartbleed (cloudflare.com)
- Heartbleed – Can CRLs cope? (cybermatters.info)
- SSL CRL activity (isc.sans.edu)
- Heartbleed certificate revocation tsunami yet to arrive (news.netcraft.com)
- OCSP Stapling in Firefox (mozilla.org)
- Internet slowed by Heartbleed identity crisis (zdnet.com)