Re: [CCAMP] AD review of draft-ietf-ccamp-assoc-ext

[Lou Berger <[e-mail]>]

Francois,

On 8/10/2012 8:07 AM, Francois Le Faucheur (flefauch) wrote:
> Lou,
> 
> On 9 Aug 2012, at 15:44, Lou Berger wrote:
> 
>> How about "MUST be" --> "is"
>>
> 
> I think the situation is that:
> 	* we want to define a Resource Sharing Association whose semantics is clearly that resources ought to be shared
> 	* at the same time, all things related to admission control, bandwidth accounting etc are out of scope, so the exact meaning of "sharing resources" is somewhat implementation specific.
> Do you agree?

precisely.

> 
> My personal preference is that :
> 	* we have a strong statement about semantics (i.e. "MUST" or a "SHOULD")
> 	* we clarify that the realization of this is somewhat implementation specific (to that end, I like Adrian's suggestion below to add an explicit statement clarifying that it may mean different things to different implementations)
> This is as opposed to an alternative approach which has a weak statement ("is") on semantics of the Association.
> 
> So I'd suggest something like:
> "Once an association is identified, resources MUST be considered as
> shared across the identified sessions by the admission control
> function. Since the admission control function is outside the scope
> of RSVP, we observe that how resource sharing is actually reflected
> may vary according to specific implementations (e.g. depending on the
> specific admission control and resource management algorithm, or on
> how local policy is taken into account). "
> 
> A SHOULD instead of a MUST can work for me to.
> 
> Would that be acceptable?

WFM

Adrian are okay with this language?

Lou

> 
> Francois
> 
> 
>> Lou
>>
>> On 8/8/2012 12:02 PM, Francois Le Faucheur (flefauch) wrote:
>>> 2205 uses the term "Admission control", so would it work if we said:
>>> "
>>> Once an association is identified, resources MUST be considered as
>>> shared across
>>>  the identified sessions by the admission control function.
>>> "
>>>
> 
> 
> Adrian Farrell wrote:
>>> actual resource allocation is really an implementation choice and
>>> different internal implementation choices.  We didn't want to overly
>>> constrain an implementation. "is expected" is really what we mean, but
>>> how do you say this in 2119 language?
>>
>> I think you are fine using SHOULD in this case. But you need to add
>> ...although implementations MAY vary this according to local policy and
>> resource-sharing capabilities.
>>
>> (Note that Francois proposed to promote this to a MUST, which would be OK, but
>> doesn't seem to be the original intent.)
> 
> 
> 
> 
> 
>