On Sun, 1 Feb 2026, Steve Crocker wrote:

The motivation for documenting the life cycle is to make both implementers and 
operators aware there are seven phases and there's a rationale for defining 
this many
phases.  Documents that give a snapshot of which algorithms are currently 
recommended, not recommended, etc. don't provide that perspective.

RFC8624 already said:

   Although this document's primary purpose is to update algorithm
   recommendations to keep DNSSEC authentication secure over time, it
   also aims to do so in such a way that DNSSEC implementations remain
   interoperable.  DNSSEC interoperability is addressed by an
   incremental introduction or deprecation of algorithms.

[...]

   It is expected that deprecation of an algorithm will be performed
   gradually in a series of updates to this document.  This provides
   time for various implementations to update their implemented
   algorithms while remaining interoperable.  Unless there are strong
   security reasons, an algorithm is expected to be downgraded from MUST
   to NOT RECOMMENDED or MAY, instead of to MUST NOT.  Similarly, an
   algorithm that has not been mentioned as mandatory-to-implement is
   expected to be introduced with a RECOMMENDED instead of a MUST.

   Since the effect of using an unknown DNSKEY algorithm is that the
   zone is treated as insecure, it is recommended that algorithms
   downgraded to NOT RECOMMENDED or lower not be used by authoritative
   nameservers and DNSSEC signers to create new DNSKEYs.  This will
   allow for deprecated algorithms to become less and less common over
   time.  Once an algorithm has reached a sufficiently low level of
   deployment, it can be marked as MUST NOT so that recursive resolvers
   can remove support for validating it.

And RFC9904 says:

  To ensure interoperability between DNS resolvers and DNS authoritative
  servers, it is necessary to specify both a set of algorithm
  implementation requirements and usage guidelines to ensure that there is
  at least one algorithm that all implementations support.

[...]

  The perspective of implementers may differ from that of an operator who
  wishes to deploy and configure DNSSEC with only the safest algorithm. As
  such, this document also adds new recommendations about which algorithms
  should be deployed regardless of implementation status. In general, it
  is expected that deployment of aging algorithms should generally be
  reduced before implementations stop supporting them.

[...]

  By the time a DNSSEC cryptographic algorithm is made mandatory to
  implement, it should already be available in most implementations.

[...]

  It is expected that the deprecation of an algorithm will be performed
  gradually. This provides time for implementations to update their
  implemented algorithms while remaining interoperable. Unless there are
  strong security reasons, an algorithm is expected to be downgraded from
  MUST to NOT RECOMMENDED or MAY, instead of directly from MUST to MUST
  NOT. Similarly, an algorithm that has not been mentioned as mandatory to
  implement is expected to be first introduced as RECOMMENDED instead of a
  MUST.Since the effect of using an unknown DNSKEY algorithm is that the
  zone is treated as insecure, it is recommended that algorithms that have
  been downgraded to NOT RECOMMENDED or lower not be used by authoritative
  nameservers and DNSSEC signers to create new DNSKEYs. This ensures that
  the use of deprecated algorithms decreases over time. Once an algorithm
  has reached a sufficiently low level of deployment, it can be marked as
  MUST NOT, so that recursive resolvers can remove support for validating
  it.

  Validating recursive resolvers are encouraged to retain support for all
  algorithms not marked as MUST NOT.


I do not understand why implementers or operators need more background
information. The RFCs above have been saying "Here is some background on
why we made these choices are carefully considerating, do things differently
at your own peril".

These RFCs also did not use the explicit lifecycles phases you are
proposing, although obviously there is a large overlap with that and
the above quoted background information on why the IANA registry looks
as it does.


Regarding the decisions as to when a particular algorithm should be moved from 
one phase to the next, I agree this is not a simple process.  The crypto 
experts will
provide guidance on when a new algorithm has passed their analysis and testing 
and when an existing algorithm should be phased out.  Those judgments cover two 
of
the six transitions, A and D.  The other transitions depend on deployment and 
use.  I didn't propose the specific means for measuring those, but I did 
include a
requirement for a designated set of experts to make those judgments.

I believe the process of using DNSOP as the group of experts
worked fine for RFC8624 and RFC9904. That is _exactly_ what DNSOP is
chartered to do. I would be concerned if we reduce that group to a few
(non-democratically) selected designated experts.

I would expect they will develop procedures, tools and criteria to do this 
work.  And to your
point in your follow-up note that people pay more attention to RFCs than 
current values in IANA registries, the decisions made by the designated experts 
at each
transition will indeed be documented in RFCs.  And likely publicized in other 
venues.

Assuming those would be BCPs or Standard Track RFCs, that requires DNSOP
activity then?

Implementers, i.e. the folks in control of software crypto packages, will make 
their own decisions, of course, but the intent here is to give them clear 
guidance,
with accompanying data, to make their decisions much easier.  Similarly, 
operators, i.e. the folks in control of which algorithms in the crypto packages 
are enabled
or disabled in their services, will also find the guidance helpful.

I think DNSOP has done that with RFC8624 and RFC9904 and I see them doing
the same for 9904bis once the time is right.

Paul

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

Reply via email to