Re: [IPFIX] RFC5815 (IPFIX MIB) IANA Consideration issues

[Juergen Quittek <[e-mail]>]

Dear all,

I support Thomas' request.

Currently, we have a registry at IANA at
http://www.iana.org/assignments/ianaipfixselector-mib

As you can see if you click the link, there the entire MIB module is
maintained by IANA. This implies that if we want to add a new selector
function, then the MIB module will be modified and implementers need replace
older versions.

This is not what was really intended by the authors of the RFC 5815 (IPFIX
MIB). The die was rather having IANA maintain a a registry for just NUMBERS
of selector functions that would be appended to OID ipfixSelectorFunctions,
as it is commonly done for smi-numbers at
http://www.iana.org/assignments/smi-numbers

So the IANA action that we need is

  - remove the registry of the IPFIX-SELECTOR-MIB
  - instead create a new entry at the smi-numbers registry
    for the prefix ipfixSelectorFunctions

There we would then add the selector functions defined in the PSAMP-MIB and
potential further MIB modules.

Please send your comments if you have questions or disagree with this
proposal. Note that if we do so, we would need to run IETF last call for the
PSAMP-MIB again, because this is a significant change to the document.
My proposal would be to discuss it on the mailing list now and have a final
discussion at the IPFIX session in Quebec.

Thanks, 

    Juergen




Am 29.06.11 13:58 schrieb "Thomas Dietz" unter <[e-mail]>:

> Dear all,
> 
> in the course of editing the PSAMP MIB draft we identified that the IANA
> Considerations in RFC 5815 (the IPFIX MIB RFC) do not make sense. Putting
> the whole IPFIX SELECTOR MIB under IANA control and creating a registry
> hereof is not what we need here.
> 
> I propose to change that and only put up a registry that registers the OIDs
> of the subtrees under the ipfixSelectorFunctions OID. Each subtree -- as
> described in the RFC -- represents a Selector Function and its parameters
> (in a table). So the root of the subtree (i.e. its OID) has to be registered
> in the new registry. In addition the document where this subtree is defined
> has to be registered. The review process setup in RFC5815 can remain as it
> is.
> 
> Implementing this change needs -- according to our ADs -- an updated
> RFC5815.
> 
> Please comment on this issue. Any feedback is welcome.
> 
> Best Regards,
> 
> Thomas