I am working on the logical model for a database. I need to use a number of code tables (tables that keep typically name value pairs. I need maintain data like products, services etc).
I am wondering if I increase the abstraction and use one table to represent the name value pairs but use a category to identify each type. Is there is any value in doing this?
The advantage with this I think is consolidating the data and probably minimizing the administration
The disadvantages may be too many joins that need to be qualified by the category type. Also, I may end up having too many self-joins.
Any suggestions???Personally I like this approach because then I don't have a bunch of hash tables scattered around the database. Adding new groups of name value pairs becomes a lot easier.
I haven't found the need to perform self-joins, but yes the large number of joins to the same table tends to be a pain. But, you'd still have to have the joins regardless (just to different tables).
Honestly, I am not sure of the performance benefits. But I think the "compactness" of the solution has value.
My 2 cents, maybe only 1 cent.
Terri
No comments:
Post a Comment