Data Modeling and the Canonical Conundrum

A Forrester Research paper I read recently stoked my interest (again) in the canonical data/information model, a once hot pursuit that seems to have cooled down in recent times. In short, the paper states that the canonical information model excludes "at rest" data (legacy systems of record, packaged applications etc.) but includes information in motion (messages, service invocations, etc.). My first instinct was to disagree with this definition…

But then, I began wondering about what, exactly, is a canonical information model, and realized that this definition only serves to highlight the confusion out there around [enterprise] [canonical] [business] [data/information/object] models… starting, as you can see, with the term itself. As somebody (Confucius? African proverb?) once said: "If you don't know where you are going, any road will take you there."The theme of the research paper, by Forrester analyst Mike Gilpin, is neatly captured in the title (Canonical Information Modeling Is Key To Many Information-As-A-Service And SOA Strategies), and is hard to argue with - in fact, I would substitute "Most" for "Many" in the title. Standardized information models (whether at the business/conceptual object level or at the data/entity level) will strongly boost your efforts to lay down a reusable service-oriented information infrastructure. On the other hand, disparate data/object definitions (whether at rest or in motion) will increase complexity and cost while reducing reliability and reuse.

As with most concepts, it is less important that we describe it neatly than that we do it well, but a common definition and understanding will help more of us reach that important goal. I suspect that a good reason many of us are mired in a morass of non-standard information and data models is that we don't know what the goal is. Is it to define a common business object model, or a common data model? What's the difference? Does this include packaged applications and existing (often legacy) data stores? Does this include data "at rest" and/or "in motion"? Whose responsibility is it? Where do we begin?

Here's another way to look it: If you don't know where you are going, you are already there - but it's not necessarily a good place to be.The definition of canonical information model only serves to highlight the confusion out there around [enterprise] [canonical] [business] [data/information/object] models... As somebody (Confucius? African proverb?) once said: "If you don't know where you are going, any road will take you there."