ietf
[Top] [All Lists]

Re: Use of private OIDs in WG (standard-track) documents

2015-04-01 13:13:28
Phillip Hallam-Baker wrote:

Stephen Farrell wrote:

On 30/03/15 20:29, Nico Williams wrote:
 that
would effectively require having procedures for *early* IANA assignment
of IETF OID arcs to upcoming Internet-Drafts.  As it is I don't think we
have procedures for that,

See RFC7120 and RFC7299 (as Russ pointed out on the trans list).
We have those procedures now.

There is still, however, no real issue to be dealt with here.

I do not see the problem either.  It doesn't matter which arc the OIDs
come from.  The only two issues that matter

  (a) the OID must be unique, i.e. not be reused with a different
      meaning _in_the_same_context_.  Reuse of the OID for different
      purpses in other contexts does not matter.

  (b) the assignment ought to be permanent from the start.
      Having to "renumber" (i.e. migrate to a different codepoint
      in an installed base) is awful and must be avoided.
      So yes, a project-based early assignment process is required.



We might be making the situation worse however by insisting that IANA
issue OIDs to organizations rather than for projects. This is one of
the few instances where IANA is acting as a non-IETF registry on
behalf of the (now defunct) ITU-T.

If we had a registry for projects, people could take an OID arc for
their project. Use it in private space during development and then
transfer control to the IETF or W3C or OASIS or wherever if the
project turns into a standards track effort.
 
Huh?

Look here:

  http://www.alvestrand.no/objectid/1.3.6.1.html

While IANA does provide Enterprise/Organization based OID arcs under
private.enterprises (1.3.6.1.4.1), it also has registries for
other purposes under 1.3.6.1, and has been using it for two decades,
such as technology-related assignments under security (1.3.6.1.5).


The problem with organizational OID arcs, as originally conceived by ITU-T,
is that they had the model of an arc "lease" through their national member
body, rather than a permanent assignment -- and for premium fees (in the
ballpark of $4000/year per organization).  Unsurprisingly, this approach
was not terribly attractive.


Instead, folks got a registration under IANA's "private.enterprises" for free,
and started assigning OIDs below those arcs for stuff other than SNMP.
You avoid the ITU-T fee/lease, and pay in size/network bandwidth for
larger OIDs in (protocol) PDUs.


-Martin

<Prev in Thread] Current Thread [Next in Thread>
  • Re: Use of private OIDs in WG (standard-track) documents, Martin Rex <=