Question

Let’s say I have a table such as:

CREATE TABLE t1 (
  col1 int,
  col2 int
);

and a “metadata table” classifying these attributes:

CREATE TABLE t2 (
 attribute varchar,
 description varchar
);

So sample data might be:

t1:

col1  | col2
------+------
 4    |   6

t2:

attribute   | description
------------+--------------
 col1       |  first column
 col2       |  second column

What are the pros and cons of using this approach? It certainly makes me uncomfortable but I can’t put into words any solid objections.

Edit: it makes me uncomfortable because this design tends towards an Entity-Attribute-Value data model where everything may as well be stored in one table, i.e. the data model no longer captures as much meaning. I’m trying to quantify how much of a problem this really is.

Answer 1

It sounds like what you are describing is Entity Attribute Value (EAV) or possibly row modeling. Search on these terms. You will find that there are a few advantages and many, many disadvantages.

Storing metadata in your data model’s repository is good. Modeling a metadata repository in your schema is almost certainly unnecessary and unhelpful.