Date Columns in FACT or Dimensions

We have three tables X, Y, Z. The table structure are the same, but the data is different. These tables are linked by a KEY column. The tables have multiple date fileds DATE_FIELD1, DATE_FIELD2, DATE_FIELD3, DATE_FIELD4. The values in these dates are different. My question is, whether these date columns will be in FACT or DIMENSION tables. Please suggest.

Raj

Simply stating metadata about columns does not provide any insight as to the correct model. This is true for OLTP modeling as well as dimensional modeling. Now if you talked about the definition of these columns, how they are used by the business, or even performance measures, I might be able to help.

If table structures are same means referring to same business process. You can now start finding dimensions from these tables and also create a detail time dimension which can be used multiple times (role playing) if you have multiple dates for a business transaction.

It boils down to what do the dates mean and which entity they belong to. If the dates related to the dimension, thats where they should go. If they relate to a business event, they belong in the fact. You use FKs to the date dimension if the date needs that context (i.e. the attributes for date). Things like effective dates in a dimension usually are just date columns.