RE: netconf WG charter proposal - data model or not?

[Andy Bierman <abierman@cisco.com>]

At 06:19 PM 4/6/2003 -0700, Faye Ly wrote:
>Andy,
>
>This is to respond to your second part of the comment:
>
>>>
>>>Your statement 'overly describing something we don't do' is a very
>sharp
>>>observation!  We are suffering from lacking a very clear vision on how
>>>this configuration protocol is going to turn out without using any
>data
>>>model.  I think of the data model as the use case that needs to be
>feed
>>>into the design as the requirement. 
>
>>I disagree that we don't know what an XML data model might
>>look like.  Look at the XMLCONF examples. Look at the Junoscript
>>draft.  Several vendors are already creating XML data models
>>on their own.
>
>When you said the XMLCONF examples, do you mean the examples given in
>"CLI-based Mediation Mechanism Using XML"?  

there are examples in sec. 4.1.2, 4.2, 4.3, 4.9, and more

>As for 'several vendors are already ...'.  This is what I am afraid of,
>several vendors are already creating XML data models.  If they are doing
>so, do they inter-operate?  Aren't we going back to the original problem
>that 'each vendor has its own configuration model and thus requires
>operator to manage each individually'?  If TOS is required in one
>vendor's configuration and 'optional' in another (or worse, not
>supported), how does XMLCONF address that?  Access list configuration is
>another good example, Cisco and Juniper do this quite differently.  How
>does XMLCONF help the operator to configure this in the same fashion?

Are we going to repeat this debate?
With SNMP, we have a standard protocol plus standard and
proprietary data models.  For XML, we need a standard
protocol plus standard and proprietary data models.

We need to do the protocol work to get started.  We can
convert standard MIBs to standard XSDs to jumpstart the
work on standard data models. (smidump already does this)

There are some people who do not believe the work on 
standard data models is worth the effort.  I tend to
agree that they are overrated.  At best we will always
have a mix of standard and proprietary data models.
The amount of code (or training) reuse actually realized 
due to use of standard data models is hard to quantify.


>-faye

Andy


--
to unsubscribe send a message to xmlconf-request@ops.ietf.org with
the word 'unsubscribe' in a single line as the message text body.
archive: <http://ops.ietf.org/lists/xmlconf/>