More FM Questions

[UcfMike]

I am modeling tables from a transactional database. I am working with about 30 tables and multiple many to many joins. I've created a transactional layer in FM which mimics, the schema of the transactional database. I'm now working on the transformation layer.

When working on the transformation layer, I'm trying to denormalize the data, because its my understanding this is what you're suppose to do when modeling in FM. I'm also tyring to simply and restrict the paths to the data, so the results I receive from my reports are correct and remove the many to many relationships. I've been doing this by looking for what I call legs of tables. Tables that join together without joining to other areas of the database.

Is this the right thought process, or is this a poor way of modeling?

Why does FM want me to go through this process, how is it thinking?

 

[blom0344]

If your database contains many to many relationships between tables then there MUST be a seperate linking table for each many to many relation set.
Effectively the many-to-many relationship consists of 2 1:n relationships.
Can you be a bit clearer on your 'legs of tables' ?

Ties Blom

 

[UcfMike]

Probably not but I'll try.

Something like this.

Table A
1
to
n
Table B
1
to
n
Table C


The many to many relationships cause me problems because I'm not sure what to do with the join table. In the transformation layer, I'll usually roll it up into on of the tables it joins with.