Re: draft-irtf-nmrg-im-dm-02.txt (Informational RFC)

[Robert Moore <remoore@us.ibm.com>]

Bert,

I think this is a useful document, that definitely warrants publication.  I
have only a couple of relatively minor comments:

1. Section 3, first paragraph: "An IM can be implemented in different ways
and mapped upon different protocols; IMs are therefore protocol neutral."
In my experience it isn't quite so clear-cut.  The reality is that when a
group of people experienced in a given DM set out to define an IM, the IM
takes on the flavor of the DM.  (I'm not suggesting anything nefarious here
- more like what happens when you don't clean out your coffee-maker well
enough before using it to brew tea.)  So, for example, there are "whiffs"
of LDAP in some parts of the DMTF CIM model.

But I'm not just making a point about what people do; there are objective
differences here as well.  A figure like the one at the top of page 4
suggests that the mappings to the three IMs are equally easy,  This is
certainly not the case.  If, for example, you IM has the datatype "real" in
it, it's much easier to map it to GDMO than it is to map it to SMI.

If the scope of this document is *just* about defining the two terms IM and
DM, then maybe what's there is OK.  But to the extent that it slides over
into being a BCP, then I think it would be better if it acknowledged these
realities.

2. It seems a little odd that the RFC 2252 DM for LDAP isn't mentioned in
any of the examples, since, between the IETF and DMTF,  that's the case
where (to my knowledge) we have the most extensive set of IM-->DM mappings.

Regards,
Bob

Bob Moore
Advanced Design and Technology
Application Integration Middleware Division
IBM Software Group
+1-919-254-4436
remoore@us.ibm.com