Page 1 of 1

slowly changing dimensions

Posted: Mon Jan 02, 2006 2:45 pm
by seeta
Hi

I am new to DataStage.I am working on datastage7.5 .I want to implement slowly changing dimension tables.What I need to do to implement those tables.Do I need to create surrogate keys or do I need to do anything else.

Thanks

Posted: Mon Jan 02, 2006 3:36 pm
by kcbland
Type 1, 2, 3, or a hybrid? How familiar are you with the types and their differences? None of them require usage of surrogate keys, but they absolutely make convincing evidence of solving problems.

I suspect that you might not be totally familiar with their implementation, but the normal steps are:

1. Acquire source row and transform
2. Acquire current target row
3. Compare new source row to current target row for differences
4. If no target row exists, insert the new row (optionally assigning a surrogate key).
5. If target row exists and there's a material difference to the new row, then either perform a type3 maneuver and update the current row; for type 2 handling insert the new row and optionally mark the current row with a retiring date; for type 1 update current row with new information.

Posted: Mon Jan 02, 2006 11:26 pm
by Sreenivasulu
Hi ,

Unlike informatica you would not be having workflow for
slowly changing dimensions in datastage.

We need to design the type1,type2 and type3 dimensions on our own way which would suit the requirements.

Regards
Sreeni

Posted: Mon Jan 02, 2006 11:42 pm
by ray.wurlod
Version 8 of DataStage EE has a Slowly Changing Dimension stage. You could use that.

Unfortunately, version 8 has not yet been released. They showed it at the show in Las Vegas in November 2005, though, so some folks have been fortunate enough to see it. It's now in beta, expecting a Q2/2006 release.