RE: [NGO] Diff-Serv datamodel for discussion startpoint

["Eduardo Cardona" <e.cardona at CableLabs.com>]

Okita, Lada, Simon, et All

Let me try to go one level up around this discussion. 

Something probably worth to mention about the  Informational Model for DiffServ described in RFC3290. It directly models DiffServ as a DAG (Directed Connected Acyclic Graph) (section 3.1 RFC 3290). Thus, The DiffServ MIB DataPath (RFC 3289) precisely calls to reuse objects (Okita's elements like Markers) among different diffServ datapaths.

Perhaps the draft indirectly references this issue as it maps the MIB objects from RFC 3289. however the singularity of DAGs could be stated directly in the document to explain why the document ends up with IDREFs pointers all across the XML schema (as SNMP did with RowPointers). 


Or, 

Personally I am not sure if DAG for DiffServ was appropriated in first place. Which ended up being used in the DiffServ MIB in RFC 3289 and now netconf data models.

Specially for externally committed resources (e.g. via COPs, or Netconf itself, or other protocols) for a particular client/session. My concern is that SNMP/NetConf static provisioning model may not be usable by dynamic external assignments due its general complexity of tracking References (RowPointers , or IDREFs), and the ownership of those IDREFs: who creates those IDREFs and when those can be changed, which services, sessions are linked to those references.


Contrary, dynamic external signaled QOS policies, are per session in some network elements, for example, access networks where each access device can be seen as a sub-interface  - Simon's "Deep"  model.

Particularly, NetConf DiffServ data Model should provide status of the difserv "datapaths" from any configuration source. 
Therefore for the specifics of DiffServ for NetConf I would be inclined for a "Deep" model for DiffServ instead of the DAG principle implemented in the SNMP DiffServ MIB - RFC 3289. 

I agree with Lada's comments that References are used in general. Nonetheless, NGO may provide best practices recommendations at some point in the future of when to use references vs. children within an XML element.

Thanks

Eduardo


-----Original Message-----
From: Ladislav Lhotka [mailto:lhotka at cesnet.cz] 
Sent: Monday, March 19, 2007 10:52 AM
To: ngo at ietf.org
Subject: Re: [NGO] Diff-Serv datamodel for discussion startpoint

Hi Simon,

Simon Leinen píše v Po 19. 03. 2007 v 16:49 +0100:
> You claim that "When parameters of a functional element linked by
> multiple elements are changed, operators would meet trouble to change
> all elements in the configuration corresponding to the
> parameter-changed element respectively."  With XML, I think this is a
> particularily weak argument, because there are good (even
> standardized) mechanisms for manipulating multiple parts of an XML
> document.

But you don't want to have the same information being represented at
multiple places, so I think in most but trivial cases some kind of
cross-referencing is needed. However, one should do it only where
necessary since links introduce other potential inconsistencies such as
missing targets. To this end, it is much easier to allow
cross-references to real objects such as interfaces as opposed to
"made-up" entities that exist only within a particular configuration.

Lada

-- 
Ladislav Lhotka, CESNET
PGP Key ID: E74E8C0C


_______________________________________________
NGO mailing list
NGO at ietf.org
https://www1.ietf.org/mailman/listinfo/ngo



_______________________________________________
NGO mailing list
NGO at ietf.org
https://www1.ietf.org/mailman/listinfo/ngo