Re: [CCAMP] index field in draft-ietf-ccamp-ethernet-traffic-parameters

["Benoit C Tremblay" <[e-mail]>]

Hello Dimitri,

Thanks for the clarification.

I am not sure to understand why we need the BW in both the EXTENDED_CLASSTYPE and the SENDER_TSPEC. Can't the BW accounting use the information from the SENDER_TSPEC (either CIR or CIR+EIR) instead of relying on the EXTENDED_CLASSTYPE?

Regards,
Benoit

> -----Original Message-----
> From: PAPADIMITRIOU Dimitri 
> [mailto:[e-mail]] 
> Sent: June-26-09 10:26 AM
> To: Benoit C Tremblay; [e-mail]
> Subject: RE: index field in 
> draft-ietf-ccamp-ethernet-traffic-parameters
> 
>  
> Hi Benoit, 
> 
> 
> see in-line for the answers:
> 
> 
> ________________________________
> 
> 	From: Benoit C Tremblay [mailto:[e-mail]] 
> 	Sent: Friday, June 26, 2009 4:02 PM
> 	To: PAPADIMITRIOU Dimitri; [e-mail]
> 	Subject: index field in 
> draft-ietf-ccamp-ethernet-traffic-parameters
> 	
> 	
> 
> 	Hello Dimitri, 
> 
> 	I have further comments on the section decribing the 
> index field. 
> 
> 	1) There is a reference to an expired draft 
> (draft-minei-diffserv-te-multi-class).  It should be removed.  
> 
> 
> I am not sure how to deal with this - remove is not an option 
> (as the doc. makes use of an object of that ref.) You will 
> find a version here 
> <http://tools.ietf.org/id/draft-minei-diffserv-te-multi-class-02.txt>
> 
> 	2) The requirement that "the index field value MUST 
> correspond to at least one of the index values included in 
> the CLASSTYPE object" is a bit problematic as a single 
> CLASSTYPE object can be present in a PATH message and a 
> single classtype can be represented with this object. The 
> requirement would then exclude the possibility to insert 
> multiple bandwidth profile TLVs in the SENDER_TSPEC.  I 
> suggest that this requirement should be conditional to the 
> presence of thhe CLASSTYPE object.  If a CLASSTYPE object is 
> present, then a single BW profile TLV shall be present and 
> the index value must be the CT defined in the CLASSTYPE 
> object.  If no CLASSTYPE object is present, multiple BW 
> profile TLV (up to 8) can be present and the index value 
> either represent a CT (values in [0, 7]) or a predefined set 
> of CT (values in [8, 247]).
> 
> 
> It is exactly the reason for using the EXTENDED version of 
> the CLASSTYPE. Note that the latter option does not apply 
> (cf. MUST) when used. 
> 
> Also important to remember the BW per CTi field MAY be used 
> for bandwidth accounting purposes (the BW per CTi does not 
> replace the TSPEC BW sub-TLV i.e. "The Ethernet SENDER_TSPEC 
> object MAY include more than one Ethernet Bandwidth Profile TLV").
> 
> 
> 	3) Also, it should be clarified that the 240 
> pre-defined sets of CTs are not standard.  The following text 
> is proposed to be included :
> 
> I will the first sentence, the other should be included in 
> i-d's making use of these traffic parameters.
> 
> Thanks,
> -dimitri.
> 
> 	"To ensure coherent operation, the network 
> administrator MUST configure exactly the same pre-defined CT 
> sets on all LSRs of the GMPLS domain. Proper configuration 
> needs also to be agreed between administrative domains if 
> this feature is used over UNI. The mean to pre-configured 
> those sets and their definitions are out of scope of this 
> specification. "
> 
> 	Regards, 
> 
> 	Benoit 
> 
> 	Picture (Metafile) 
> 	Benoit Tremblay 
> 	Broadband and Systems Research 
> 	Ericsson Canada 
> 	8400 Décarie    Mount-Royal 
> 	Tel: [phone] x45064 
> 	Cell.: [phone] 
> 	Fax: [phone] 
> 	mailto: [e-mail] 
> 
> 
>