Reed Me

Development schtuff, SQL Server & other random geekness. Now with more [fill in the blank]!

Do we need a new term? Metametadata?

Do we need a new term? Metametadata?

  • Comments 1

Malcolm's got a good perspective that doesn't get discussed often enough by practitioners of the black arts (data modeling, database design, database architecture, whatever -- datamancy? dataturgy? gross!).

Why Data Models Do Not Work: The Role of Reference Data (TDAN.com) (With thanks to Steve @ SQLServerCentral for the link.)

I've seen meeting-type exercises in which the users of overnormalized (heresy!) data models find it difficult to have any conversation about the model and the data it allegedly represents above the very basic without resorting to a T-SQL-like syntax. I'm not suggesting that logical data models don't have value; I'm just suggesting that where they dramatically differ from the physical, their value is diminished to everybody except the person(s) who gets paid to keep them confusing.

For example, if all of your "master data" is hypothetically abstracted into a database containing only tables for Entities, Attributes, Hierarchies and Relationships, it becomes really difficult to discuss the taxonomy of your enterprise data model (customers, business units, oil wells, whatever) using the logical data model that doesn't have any of those things in it. [I'm not holding you entirely to blame, Jamie, so don't take this too personally!! Heh.]

Reference data is the metadata, right? So it might be easier to say "data model" than metametadata and it probably makes more sense, but what's the fun in that?

Leave a Comment
  • Please add 2 and 1 and type the answer here:
  • Post