I am the assigned Gen-ART reviewer for this draft. The General Area
Review Team (Gen-ART) reviews all IETF documents being processed
by the IESG for the IETF Chair. Please treat these comments just
like any other last call comments.
For more information, please see the FAQ at
<http://wiki.tools.ietf.org/area/gen/trac/wiki/GenArtfaq>.
Document: draft-ietf-eppext-keyrelay-10
Reviewer: Robert Sparks
Review Date: 25Nov2015
IETF LC End Date: 4Dec2015
IESG Telechat date: (not yet scheduled)
Summary: Ready for publication as Proposed Standard with nits
This is a small nit, but please consider changing the document to
address it. The motivation for this extension leans on improving the
security of transferring information between registrars. It should be
recast as providing better automation and reliability instead. In
practice (and I think in specification), it hinges on passing a password
from the registrar of record to the gaining registrar through some
unspecified means (though typically through the registrant). That
password is required to be placed in the create by the gaining registrar
as specified in this document in order for that create to succeed at the
registry. While it would be impractical and error-prone, the same
channel that was used to hand this password around _could_ be used to
pass the keying material this extension addresses.
Reading draft-koch-dnsop-operator-change (an informational reference
currently) helped greatly with understanding this document. That draft
expired in 2014. Please be sure it advances, and consider making it a
normative reference.
If it is not going to move forward, consider pulling some of the
transfer mechanic recommendations and the definitions of losing/gaining
entities into this draft, unless they've already made it into the RFC
series somewhere else?
The security considerations document says a server SHOULD NOT perform
any transformation on data under server management when processing a
<keyrelay:create> command. Can this point to more detailed discussion
somewhere? Why is this not a MUST NOT? (What are the conditions where
violating the SHOULD NOT is the right thing to do? What are the risks a
server takes if it performs such a transformation?)
Micro-nit : In section 2.1 where you say "The <expiry> element MUST
contain one of the following", consider saying "The <expiry> element
MUST contain exactly one of the following".
RjS