Re: [IPFIX] draft-ietf-ipfix-ie-doctors-04.txt

Hi, Paul,

Interesting. 

I hadn't checked the registry, but I can say that the FCFS policy for Units was intentional: imagine a an enterprise-specific information element designed to do something very odd with unorthodox units, say, counting rats crawling around in the underfloor space in a data center (perhaps to attempt to correlate packet loss with rodent infestation). Now, the IEs for this would be ESIEs, but since this is a research study, we'd want to annotate the data (stored in files) with 5610 type information records. 

However, on review, perhaps Units in the IANA registry should cover only those IEs in the IANA registry, and if you want to do something really odd, you should leave units blank and explain "expressed in rats per cubic meter as measured by the Yoyodyne Underfloor Rat Counter model 19" in the informationElementDescription field. But that was the intent.

If we decide this, the proper thing to do is to go with the registry as is and change IE-DOCTORS to update RFC 5610.

Cheers,

Brian

On Sep 3, 2012, at 5:11 PM, Paul Aitken wrote:

> Brian, All,
> 
> There's a disconnect between RFC 5610 and IANA's IPFIX registry which affects ie-doctors:
> 
> Section 3.7 of RFC 5610 defines IE units and states, "new types may be added on a First Come First Served [RFC5226]     basis."
> 
> Cection 4.4. of IE-doctors confirms: "Note that the Units subregistry is maintained on a First Come, First Served basis."
> 
> However IANA's registry states that Expert Review is required. See http://www.iana.org/assignments/ipfix/ipfix.xml#ipfix-information-element-units
> 
> 
> I suggest that Expert Review is more correct, since FCFS has no review. From RFC 5226:
>             There is no substantive
>             review of the request, other than to ensure that it is
>             well-formed and doesn't duplicate an existing assignment.
> 
> 
> Currently anyone could request any new units, and IANA would have to honour the request even if it was stupid or irrelevant.
> Whereas with Expert Review, stupid requests could be rejected since "approval by a Designated Expert is required."
> 
> 
> Regarding Expert Review, RFC 5226 says:
>             The required documentation and review
>             criteria for use by the Designated Expert should be provided
>             when defining the registry.  For example, see Sections 6 and
>             7.2 in [RFC3748].
> 
> 
> Currently the "IPFIX Information Elements" and "IPFIX Information Element Units" require Expert Review without defining any documentation or review criteria.
> 
> We should address that, eg, "Experts must double check the ...whatever... with already defined label types for completeness, accuracy, relevance, and redundancy."
> 
> P.
> 
> 
> 

_______________________________________________
IPFIX mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/ipfix