[OPS-AREA] Converting SNMP MIBs to SOA/Web Services Management Artifacts

Hi Andy,

I've started a new thread on this topic raised by your comment on the
previous thread:
"I am not convinced that constraining any new high-level API for the
sake of SNMP Proxy would be a popular decision."

To be clear about the primary objectives I have in mind for the
proposed effort:

- To make the existing (and future) body of SNMP MIB data models, and
consequently the instrumentation implementing them, more readily usable
by the emerging breed of SOA/Web Services management tools (what you
referred to as "SOA/WSDL NMS applications" in the earlier thread).

- Making the data models (in the form of XML, RDF, WSDL, SML, CML,
WS-Policy, OWL, etc...TBD by the proposed effort) available to those
management applications has value in and of itself in a number of ways,
such as:
   - Easier to form a unified view of services and the infrastructure
supporting them
   - Easier to articulate, understand, manipulate, and exploit semantic
relationships among those entities
   - Elimination of the need to understand/support ASN.1 or even the
SMIv2 directly

- Making the existing (and future) SNMP instrumentation available to
those tools -- transparently to the tools wrt SNMP and SMIv2 themselves
-- is a huge gain for:
   - The user community, through the elimination of the need for
separate management applications for services versus infrastructure
management.
   - The SOA/Web Services Management tool vendors, through the
relatively facile expansion of managed resources which will be
available to them (at comparatively little incremental development
cost).
   - The managed entity suppliers, by extending the useful lifetime of
those products (without requiring new instrumentation on those
products).
- Making this SNMP instrumentation available in this context would
almost certainly rely on a proxy model, in which the proxy would use
the SOA/WS stuff on one side and handle the SNMP interactions on the
other, ideally in a many-to-many arrangement.

All of that, in my view, happens w/o developing APIs...rather by
standardizing the methodology for converting SNMP MIBs to SOA/WS
management artifacts.  That APIs can be layered on the effort to
deliver even more value is beyond question -- I just don't see it as a
prerequisite for delivering the kinds of value I outlined above.  I
would like to know your thoughts on that perspective and to better
understand your own views.

(By the way, I personally have a huge user community in mind that needs
this solution...but I suspect it will have very broad appeal.  Current
SNMP-only management application vendors might feel a bit threatened by
it, but I think the writing is on the wall on that one...and they might
be prime candidates to provide the proxy entities, if they want to
compete in that space.  Disclaimer: I personally no longer have any
involvement in the management product business, of any kind -- I am
strictly a trusted advisor to a large user community.)

Cheers,
BobN

_______________________________________________
OPS-AREA mailing list
OPS-AREA@ietf.org
https://www1.ietf.org/mailman/listinfo/ops-area

[OPS-AREA] Converting SNMP MIBs to SOA/Web Services Management Artifacts - API vs Data Models