- Ref: DataReferenceModel_09_2004 (2KH5)
- Comment: Volume Strategy and Relationship to ISO 11179 (2KH6)
- Volume I – Data Reference Model (2KH7)
- Information Reference Model – The DRM should be the IRM, an Information Reference Model. A data-centric architecture is anachronistic. For three decades, the industry has been shifting its emphasis from data to information. Technologies for the formal description of information have emerged and matured in the last decade or two. (2KH8)
- An approach similar to ISO/IEC 11179 is being undertaken by ISO 19763, currently in working draft state. This work-in-progress is working out a registration and categorization framework for ontologies (formalized descriptions of semantics on which the concept of information is predicated). If the FEA intends to be both current and forward-looking, it needs to examine this and similar efforts to use as a basis for an Information Reference Model. (2KH9)
- Volume II – Exchange of Data (2KHA)
- The mission of this volume is orthogonal and ancillary to the DRM's primary mission of data classification. (2KHB)
- Information messaging is a complex topic with numerous nuances. This is a "slippery slope" that will take the DRM from the level of abstraction generally (and appropriately) addressed by the FEA to a greatly detailed specification that will be overly tied to implementation and therefore destined for early obsolescence. Additionally, the point is not "data exchange", but rather "Information Exchange", this can be done by sending documents of diverse formats, binary messaging, and can employ numerous gateway technologies. It is not clear to me that there is benefit in addressing it in this volume. (2KHC)
- While the DRM asserts that it is using information interchange concepts from ISO/IEC 11179, that standard wisely keeps this issue at a distance while acknowledging it as a related concept, noting only its applicability to interchange. (2KHD)
- Over-arching Model of All Things Federal (2KHE)
- ISO/IEC 11179 is a reasonable enough standard for categorizing data, building on industry experiences such as AFI 33-110 (formerly AFR 4-29). In spite of national and international efforts to categorize data for re-use, practical success has remained spotty, confined, and modest. Among the reasons for this is a lack of partitioning and segmenting in the aggregation of data, and providing effective projection of categorization for searching. (2KHF)
- Enterprise vs. Solution Architecture (2KHG)
- There are two common metaphors for systems architecture: (2KHH)
- 1. Building Architecture – by describing different aspects of a system, the system is specified sufficiently to build it. In the metaphor, one specifies the land plat, foundation, frame, typical frame corner construction, plumbing, electrical, etc… and the relationships among all of the aspects. This metaphor is directly applicable to "Solution Architecture" where applications or a family of applications are designed in the context of their target environment(s). (2KHI)
- 2. City Planning – by describing different environmental aspects, a complex environment of environments can be specified through a relatively simple set of "building codes", "zoning", and "city planning". This metaphor is directly applicable to "Enterprise Architecture" where a complex environment and the inter-relationships of the sub-environments are specified, allowing the panoply of environments to be populated by applications which will cooperate and inter-operate where they need to. The Enterprise Architecture provides the constraints within which Solution Architects operate. (2KHJ)
- It is difficult to tell from DRM Volume I by itself, but it appears that there is a danger that the DRM may attempt in practice to provide a detailed "Solution Architecture" type approach to all data in the Federal Government. I think the DRM should focus, like the rest of the FEA, on the higher level "City Planning" paradigm. (2KHK)
- An approach that provides flexibility within an Agency or department within and agency would be preferred. When information is needed among multiple agencies, there should be a searchable information catalog (as the DRM suggests), but also facility for establishing protocols for collaboration without constraining the technology that implements the protocol. (2KHL)
- Data Warehousing (2KHM)
- Modern search engine technology may be enough to facilitate the location of shared data elements, but it may be that data warehousing principles and organization need to be applied. ISO/IEC 11179 provides a start for some of the information needed, but there may need to be additional specification of data warehousing architecture in order to assure that the catalogs of diverse agencies can all be searched effectively. Cataloging of data or ontologies will need to be a distributed activity. This activity will need to be sufficiently specified in order to achieve interoperability in catalog search tools. Consequently, rules and "boundary conditions" for cataloging and organizations may need to be strengthened beyond that provided by ISO/IEC 11179 or the fledgling ISO 19763. (2KHN)
- ISO/IEC 11179 Interpretation and Commitment (2KHO)
- On the assumption that the FEA should actually have a data-oriented reference model, factors related to the draft DRM's interpretation and simplifications of ISO/IEC 11179 have made the DRM ambiguous and self-conflicting. (2KHP)
- Commitment to ISO/IEC 11179 – If there must be a cataloging data-oriented architecture, it would be best to commit to ISO/IEC 11179. Though there may be some weaknesses to 11179, it is an International Standard with an established forum for evolution and correction. Examples of the DRM's weak commitment to the standard are (emphases added): (2KHQ)
- Simplifications of ISO/IEC 11179 have Introduced Inconsistency – Efforts to simplify concepts in ISO/IEC 11179 for over-arching discussion have created an imprecise conceptual "landscape" that is surely not that which is intended. For example the description of "Structure of Data" on page 11 suggests 1:1 cardinality of "Data Object" relationships with a "Data Property" and a "Data Representation", making the concepts unworkable. There are numerous other examples. The DRM should carefully align its presentation narrative with the semantics of the standard. (2KHV)
- Unnecessary introductions of and changes to the vocabulary of ISO/IEC 11179 – These changes result in terms being used in a way that is inconsistent with industry-wide customs of usage. This introduces unnecessary confusion. Some examples: (2KHW)
- e. "Data Object" is used where the term "Object Class" (as used in ISO/IEC 11179) would be more appropriate. The term "Data Object" usually implies the concept of an instance rather than a class of instances; the latter being surmised as what is intended given the context of its use. (2KHX)
- f. "Super Type" is a widely used term in architecture (esp. object oriented technology) for class generalization and abstraction. A more intrinsically generic term such as the term "conceptual category" (used in the DRM's definition of "super type") would invoke less implied meaning. An even better approach would be to return to the vocabulary and concepts of ISO 11179-3. (2KHY)
- Information Exchange Package – The DRM notes that The information exchange package represents a set of data that is transmitted for a specific business purpose. It makes use of the ISO/IEC 11179 concept of Information Interchange. (2KHZ)
- g. ISO/IEC 11179 makes occasional suggestions that it can be applied to such a concept, but it is a peripheral topic without formal semantics. (2KI0)
- h. The DRM does not formalize the semantics and probably should not. Information messaging is a complex topic with numerous nuances. This is a "slippery slope" that will take the DRM from the level of abstraction generally addressed by the FEA to a greatly detailed specification that will be overly tied to implementation and therefore destined for early obsolescence. ISO/IEC 11179 wisely keeps this issue at a distance while acknowledging it as a related concept. (2KI1)