Mary Poppins was on my mind the other day. The lyrics of Spercalifragilisticexpialidocious were an ear worm that stayed with me most of the day. During this memory rewind back to the 60s, it occurred to me that I did not know the meaning of supercalifragilisticexpialidocious. Off to Google I went.
An extremely long word, used in an annoying song that has no real meaning. – Urban Dictionary
Searching for the meaning of a word is a common task for data modelers. It is at the core of what we do in our profession. We must understand the meaning, usage and characteristics of data to design our data models. When a word like supercalifragilisticexpialidocious comes along, we put on our metadata hats.
Data governance has gained considerable support over the years. It would be nice to turn over much of our metadata duties to data stewards and the data governance process. I am happy to report that my business partners do a good job in the definition and governance of their business terms. Modelers know that a data model is more than business terms. Let’s look at some of the metadata gaps seen most commonly in data models.
- Undefined relationships defeat the purpose of creating a data model. Relationships are the business rules that bring our data to life. Whenever you hide a relationship’s text or fail to enter the text. You do your model a disservice.
- Self-defining definitions are just as useless as no definitions. “Customer name is the name of the customer.” Tells me nothing about why the customer name is in this entity. State the business need for the attribute. That’s what your audience needs to know.
- Missing definition of non-business attributes such as audit and versioning fields and primary, foreign and alternate keys are dead-ends the model. DBAs and developers need this metadata to connect the logical and physical data models.
- Poor and poorly enforce naming confuses everyone. Business names are governed in our glossary. Modelers may not be the best name managers in our domain. Our job is to be consultive and thoughtful where standards and reality collide.
- No enterprise alignment makes the job at hand easy but costs in the long run. Often due diligence is not performed in the sake of saving time and resources. Always remember that metadata exists is to align like objects in the enterprise to a single definition.
- Overdosing on metadata may be the wrong answer. Thorough documentation is always a desired state. It is easy to over document metadata. Caution should be exercised when including valid values, software product references and such that quickly become obsolete.
- Underexposing data model metadata to the masses is a failure for data modelers. If the value of data modeling is only creating pretty pictures, the value of viewing the underlying metadata is undersold.
That brings me back to Supercalifragilisticexpialidocious. I knew that it had some relationship to data modeling and metadata. It is in plain view in the opening lyrics. How insightful of the folks at Walt Disney Company.
When trying to express oneself, it’s frankly quite absurd,
To leaf through lengthy lexicons to find the perfect word.
A little spontaniaty keeps conversation keen,
You need to find a way to say, precisely what you mean…
Sorry about that ear worm. I at least hope it put you in a happier place today.
I am presenting at Enterprise Data World 2014 in Austin , Texas. I hope that you will join me in my session: And Other Duties As Assigned – Embracing New Roles to Grow in Your Enterprise. #EDW14