My take on this draft:

TLDR: it increases complexity for too little gain

RFC 4034 reserves two algorithms for private use (253 and 254). I have no
desire to support private algorithms, and I think most people agree because
since the publication of RFC in March 2005, these features have seen
hardly any use. It is telling that the draft to fix the DS issue comes
20 years later. The obvious thing to do is to mark this feature as
deprecated and move on.

There is however the issue of experimenting with PQC algorithms and assigning
code points to those algorithms. Algorithms 253 and 254 seem in an ideal
position to support those experiments. Note the word 'experiments'.
As far as I know there is no proposal to run production traffic using
algorithms 253 and 254. If a PQC algorithm is deemed suitable for 
production traffic then we should allocate a code point.

A direct implementation of the current DNSSEC specification may lead to
DNSSEC validation errors when algorithms 253 and 254 are used. Validation
errors can be avoid but then may lead to downgrade attacks. In short, 
draft-andrews-ds-support-for-private-algorithms fixes a real protocol issue.

>From a software point of view, it is desirable to support experiments
with PQC with minimal impact on the core of the DNSSEC components.

At the moment (in our software) support for algorithms 253 and 254 does
not exist, but can be added easily in the adaption layer between the
generic DNSSEC code and the underlying crypto libraries. Support for
new hash algorithms, for example SHA3, would end up in the same place.

This draft changes all of that. Support for SHA-256-PRIVATE would require
generic DNSSEC code to understand algorithms 253 and 254 or hacks
in the adaptions layer. Neither of which are desirable for an experiment.

In sort, this draft fixes real issues with the current specification of
DS for algorithms 253 and 254. However, if the main use of those
algorithms is expected to be experiments with PQC, then the occasional 
DNSSEC validation error would not invalidate the experiment.

So in my opinion it would be best to abandon this draft.


_______________________________________________
DNSOP mailing list -- [email protected]
To unsubscribe send an email to [email protected]

Reply via email to