[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]
Subject: [dcml-frame] Ontology and other DCML spec questions
The first 3 of my "big Questions", on reading the 1.3 spec,
relate to the Ontology questions raised by Fred Cummins. The last 3 could be
viewed as giving DCML tooling teeth (as also raised on the last call). - The spec mentions OWL Full as the DCML ontology language dialect. While
a processor could internally use OWL Full, this choice seems too general for an
interchange format. What were the reasons for not restricting to a more
tractable sub-set? Examples of when the full power of OWL is needed/useful
would help. RDF (up to OWL Lite) would be more tractable and interoperable and
amenable to automation? - The ability of semantic Web technologies to effect translations to
other concrete data representations and languages is mentioned in the spec
several times but no examples are given. Would this be a way to address some of
Fred's issues? We would provide both ontology for model definition and exchange
and specify (automated) mappings to/from "object oriented" data
models and schemas. - In particular, the somewhat simplistic inclusion of CIM (as blobs of
data) seems a lost opportunity. Could we (the standard) define an automated mapping
(using RDF) to make CIM data available to DCML processors or allow DCML
processors to manipulate the world using CIM/WBEM? (please see below) - The rules section of the spec is currently AWOL.
Seems to me that this may be some of the "teeth" that Fred identified
as missing. Here "Policy" specification would be a desirable goal,
rather than implementation level rules. If DCML had a policy/rule language specified,
that manipulated both the data center model and (through the model) the real
world, then its usefulness would be increased tremendously. - Events seems critical to data center manipulation.
Should we include a template section for event types, event routing / transform
"deployment"? - Deployed instrumentation and effectors could also
be described and, like events, may need to be linked into DCML rules. Would "bindings"
to WSDML, JMX or other management fabrics help, as this would put DCML directly
in the data center control loop? Regards, Andre PS The OASIS Symposium is April 24-29, |
[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]