I have read much of rfc5706bis, and Richard's complaint.
I have some sympathy for whether or not it's within opsawg's mandate; however
if the AD and the IESG say it's the place to do this, then it is.

I was around when RFC2223 and RFC3552 were created.
I note how trivial the text in RFC2223 is (there will be a secton says
Postel), but also note that RFC3552 was needed to clarify, and doing so was
an IAB act. (With EKR a member, and acting, I think, as editor of the
document).  If this document came from the IAB, I wonder if people would feel
differently.

Toerless Eckert <[email protected]> wrote:
    > Sec: I do agree that security considerations are often? not as good as
    > they should be. Maybe SEC AD should more often ask "have co-authors
    > read rfc3552 and applied it to the security considerations section".
    > But i am also not sure if rfc3552 is the best it could be after now
    > being 22 years old. But i need some more time (independent of this
    > rfc5706bis effort) to have more well founded technical opinion about

In some documents, Security Considerations is the first anyone thinks about
security.  That's a flaw.  SC should review the threats, and point out the
mitigations that are aleady in the protocol description.  No BCP14 language
belongs there, because it SHOULD already be elsewhere.
Having said that, there are many protocols which are mere extensions of
others, and there really is little more to be said.   That's okay: the poit
of the mandatory sections is to have people think about the topic.

    > Mandatory Ops/Mgmt section: I also don't think that a simple "all IETF
    > proocol specification MUST have an Ops/Mgmt section" to be the best
    > possible outcome of the WG process we are starting now. I think the
    > starting point really is "all IETF protocol specification MUST have
    > Ops/Mgmt documentation".

I think that the document explains itself well enough when it says:

  Retrofitting operations and management mechanisms is
  often hard and architecturally **unpleasant**

while I would not insist that a YANG model be created for every protocol, I
think that the activity of thinking about what things need to go into that is
important.  And I think that the document goes to some length to say exactly 
that.
(IPsec, for instance, has a "PAD", which I find obtuse to talk/reason
about, as there aren't really any sensible primary keys)

I think that it's acceptable for "Operations Considerations" to say nothing,
and that option is very clearly included in section 3.2.

    > The fact that security sections turn out to be less good than they
    > could be in cases is, i think also a good reference for attempting to
    > improve on our Ops/Mgmt documentation:

I find about half the Security Considerations sections to be inadequate,
and of those I find many to be "harmless but useless" (a term I learned from 
Adrian).

I got bored/lost-interest reading rfc5706bis around section 5.
If I have any complaint, this document could probably be 1/3 in size.

--
Michael Richardson <[email protected]>, Sandelman Software Works
 -= IPv6 IoT consulting =-                      *I*LIKE*TRAINS*



Attachment: signature.asc
Description: PGP signature

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

Reply via email to