logoalt Hacker News

teddyhtoday at 3:34 AM0 repliesview on HN

> First, they require changing the operating system

This was done very quickly with IPv6; most major vendors had adequate support very early. This shows that it can be done when the companies involved actually want to do it.

> IPv6 requires you not only to update your OS

Blatantly false. AFAIK, all mainstream OSs today has enough IPv6 support to work adequately in a theoretical IPv6-only environment.

> Even the WebPKI has somewhat better NTT properties than DNSSEC: you can get a certificate for any domain you own without waiting for your TLD to start signing (admittedly less of an issue now, but we're well into path dependency).

Wait for CDS and CDNSKEY record support to be more widespread among TLDs (some support it today, and from what I can see, the number is increasing). Then you don’t need even your registrar to be involved in you DNSSEC deployment, you can just enable DNSSEC in your DNS server and let it deploy automatically.

> being incrementally deployable is a huge part of good design.

Oh, agreed.

> [0] DNSSEC doesn't strictly require this, but if you want it to integrate with IKE, it does.

Yes, this kind of new feature would have to be implemented in a backwards compatible way, with fallback to normal connections if the other end does not support it. One idea would be to put KEY records in the reverse lookup zones; only if such a record exists will you get automatic IPsec.