[Cryptography] Against against DNS (Re: New SSL/TLS certs to each live no longer than 47) days by 2029

[David Conrad]

Peter,

On Apr 24, 2025, at 11:22 PM, Peter Gutmann <pgut001 at cs.auckland.ac.nz> wrote:
> It actually stretches over a much longer time period, Thomas Ptacek's original
> series of essays, which went into much more detail than the 2015 post, was
> "The Case Against DNSSEC" from 2007, about the same time attempts were first
> made to deploy it.  

The root of the DNS wasn’t signed until 2010 and DNSSEC+DANE wasn’t standardized until 2015. Thomas’ arguments against deploying DNSSEC may have made a lot more sense prior to these events, but the validity of those arguments have been decreasing over time to the point now they feel a bit like “get off my lawn”. 

> The APNIC post, incidentally on a blog run by an organisation charged with deploying DNSSEC, 

1) APNIC, an IP address registry, has essentially no role in deploying DNSSEC. They manage the in-addr.arpa and ip6.arpa zones for the reverse mapping of IP address blocks they receive from IANA. Those mappings are DNSSEC-signed, btw. However, few people (other than email server operators) care about these mappings. APNIC is supportive of DNSSEC deployment, but not involved.

2) Geoff Huston states in his blogs at APNIC labs: "The views expressed by the authors of this blog are their own and do not necessarily reflect the views of APNIC.” Knowing both Geoff and the folks at APNIC, I doubt Geoff bothers to even run his blogs by any corporate gatekeepers prior to posting.

> that it's essentially failed, is telling:

This is like arguing IPv6 has failed (which Geoff has also argued — he likes being provocative). Depends on your point of view.

> It's solving a problem that most people don't care about at a cost that most people do care about.


Sure. As has been shown time after time, most people don’t care about security, particularly if they have to pay for it or do something different.  That is, until they get bitten.

DNSSEC+DANE can solve essentially the same problem as TLS but in an arguably more secure way that few people understand or care about (i.e., a single trust anchor vs. a forrest with transitive trust). End users aren’t going to demand this and historically, browser vendors argued the additional lookups DNSSEC+DANE implied had a negative impact on web page viewing times, so they had no interest in adding the code to support DNSSEC+DANE. Any increased security represented by the different trust model was never considered important enough by the browser vendors (for whatever reason) to outweigh the costs of deployment and the other use cases for the security represented by TLS/DNSSEC+DANE fall into the noise.

Regards,
-drc

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 322 bytes
Desc: OpenPGP digital signature
URL: <https://www.metzdowd.com/pipermail/cryptography/attachments/20250425/547d9e0c/attachment.sig>