Merkle Tree Certificates

(datatracker.ietf.org)

38 points | by mooreds 2 days ago ago

13 comments

  • nickf a day ago ago

    The lede buried here (or maybe not here but in the root program policies for MTCs) is that they’ll be 7 day maximum lifetime. And ACME only. And potentially the only thing Chrome supports (meaning you have to) by mid-2028.

    If you don’t have your public TLS automated today, better get started…

    • davidben a day ago ago

      Chrome’s policy has a maximum of 47 days, not 7 days. It’s the same max lifetime from the CA/B Forum ballot a couple years ago: https://googlechrome.github.io/chromerootprogram/cqrp/draft-... https://cabforum.org/2025/04/11/ballot-sc081v3-introduce-sch...

      There is an Apple announcement that says 7 days, but they haven’t posted a draft policy yet and says they’ll take feedback on the policy in general. https://groups.google.com/a/chromium.org/g/mtcs/c/6FPUxIiAdh...

      But, yes, automating your certificate renewals is definitely a good idea.

      • nickf 7 hours ago ago

        We know each other by proxy (multi-decade colleague of Rob, and long-time botherer of your colleagues - and with my name on that 47 day ballot ;)

        I’m reasonably certain Apple aren’t going to change things in their draft (and that’s after feedback given), but we’ll see. I’d just be glad if everyone got their automation in place already, and I think a 7-day lifetime is the forcing function for that. Not that anyone is practically prepared for it, but hey!

    • cpach 19 hours ago ago

      I would assume ACME (the way you request the cert) is orthogonal to the type of the cert.

  • ggm 2 days ago ago

    Fixes/Mitigates a huge problem with OCSP and CRL scaling in a world where every website has a cert.

    Shorter lifetimes are good. Kind of an antipattern when you thought you got a cert forever, (like 2+ years) and it turns out getting them short lived is just better for everyone.

    • cryptonector a day ago ago

      Shorter lifetimes are the better revocation mechanism (no need for revocation, just wait a bit). This I-D is about optimizing PQC and combining in certificate transparency (CT).

  • ajasmin a day ago ago

    Given I know what X.509 certificate chain and Merkle tree are.

    Can someone give a cliff note version of what this RFC brings to the table?

    • cryptonector a day ago ago

      It's in the abstract:

         This document describes Merkle Tree certificates, a new form of X.509
         certificates which integrate public logging of the certificate, in
         the style of Certificate Transparency.  The integrated design reduces
         logging overhead in the face of both shorter-lived certificates and
         large post-quantum signature algorithms, while still achieving
         comparable security properties to existing X.509 constructions and
         Certificate Transparency.  Merkle Tree certificates additionally
         admit an optional size optimization that avoids signatures
         altogether, at the cost of only applying to up-to-date relying
         parties and older certificates.
      
      IOW: it's an optimization for PQC signature algorithms.
      • Zecc 6 hours ago ago

        That didn't help me much to be honest. I much preferred reading the first paragraph of chapter 3, Overview:

            In Certificate Transparency, a CA first certifies information by
           signing it, then submits the resulting certificate (or
           precertificate) to logs for logging.  Merkle Tree Certificates invert
           this process: the CA certifies information by logging it, then
           submits the log to cosigners to verify log operation.  A certificate
           is assembled from the result and proves the information is in the
           CA's log.
        
        I didn't read more than that, as I lost interest. But at least I felt my curiosity was satisfied.
    • pseudohadamard 8 hours ago ago

      Yet another massive amount of complexity, bloat, re-engineering, and most of all attack surface dreamed up by people with nothing better to do with their time (I've half-slept through some very long talks on this stuff). In the meantime we've got daily reports of AI-assisted 0day in everything around us which is where the real problems lie.

      [pause]

      Yeah, I think that's about all you need to know. Exactly zero real-world attacks will be stopped by this, but a lot more attacks will be enabled by design and implementation flaws in it, as well as the destructive interaction of multiple incompatible mechanisms, the existing stuff and the new Merkle-tree stuff.

  • _hyn3 2 days ago ago

    This looks fantastic. (Huge fan of Merkle trees and this is a great use for them)

  • ranger_danger 2 days ago ago

    So, it's like a centralized blockchain of sorts?

    • chews a day ago ago

      POA network... but ya