RE: question on PIB and Counter32/Counter64

["Romascanu, Dan (Dan)" <dromasca@avaya.com>]

Bert,

Can you explain your concern? Is this about whether the additional semantics is needed? Should we not rely on the judgment of the authors? Assuming it is, do you have a proposal for an alternate way of providing it? Also, is there a problem to use these TCs in MIBs? 

Thanks,

Dan


> -----Original Message-----
> From: Wijnen, Bert (Bert) [mailto:bwijnen@lucent.com]
> Sent: Thursday, December 19, 2002 4:30 PM
> To: Mreview (E-mail)
> Subject: question on PIB and Counter32/Counter64
> 
> 
> If you could take a quick look at 
> 
>   
> http://www.ietf.org/internet-drafts/draft-ietf-rap-feedback-fr
> -pib-04.txt
> 
> section 3.2, then you can read:
> 
> 
>    The SPPI does not support the Counter32/64 textual 
> conventions (TC) 
>    of SNMP - for feedback collection two similar textual conventions 
>    have been defined in this PIB: Usage32 and Usage64. 
>  
>    In addition to the differential functionality of 'Counter', where 
>    only the difference between two samples generally carries 
>    information, a single value of a 'Usage' attribute usually 
> provides 
>    absolute information, since 
>    - its initial value is known (0) 
>    - no wrap-around events should occur 
>    - the time or event when the initial value was set should be    
>      available directly or indirectly from other objects. 
>     
>    When 'Usage' attributes are defined in a PRC, events that could 
>    cause a reset of the attribute to it's initial value should be 
>    defined in the description as well as the mechanism that 
> allows the 
>    PDP to detect the time of the last reset. 
>     
>    No usual COPS activity however should cause the reset of a Usage 
>    attribute. In the case of a suspension of monitoring activity 
>    (frwkFeedbackActionIndicator set to 
> 'suspendMonitoringAndReports'), 
>    'Usage' attributes should keep their values and continue counting 
>    after monitoring is resumed. 
>     
> 
> I don't agree with that and I worry about them defining Usage32 
> and Usage64 TCs to be used instead of Counter32/64 and/or
> ZeroBasedCounter32/64
> 
> Any comments?
> 
> Thanks,
> Bert 
> 
>