DMN Demystified, Part 8. What's Not in DMN 1.1

DMN 1.1 is the first implementable version of the standard. The spec is done, but won't be released to the public until mid-2016. A lot of work went into it, so people are surprised at all the things that are not in it:

  • Business glossary. Howls of anguish went out online when it was revealed that DMN would not specify a Business glossary, a business-maintained data dictionary for decision models. Actually, all of the elements that go into the glossary - the element's name, description, datatype, allowed values, etc. - are defined in DMN. What's not defined are the specific formats, metadata, user interface, versioning, and access control that one would need to use such a glossary. It would be pointless for DMN to specify these, as tools would do their own thing anyway.
  • Structured Business Vocabulary and Rules, another OMG standard, associated with Ron Ross. It was originally anticipated that DMN would link somehow to SBVR, but it didn't happen, for a variety of reasons: policy/constraint rules, not compositional, heavyweight, etc. It's a touchy subject.
  • Anything to do with methodology. A bit of FICO BKM methodology did manage to sneak in, but in general OMG tries to keep suggestions of a "right way" to apply the standard out of their specs.
  • Anything to do with testing and analysis. Completeness checking, consistency checking, test case generation, and end-to-end simulation are all important to using DMN effectively, but the spec says nothing about any of them.
  • Anything to do with execution. Actions to obtain input data, data access methods, actions triggered by the decision result, fault handling, and performance optimization are all important, but out of bounds for the DMN spec.
Some of this may make into DMN 1.2, but most are areas that OMG likes to leave open so that tools and consultants can provide their "value add."