Changing Data in the Warehouse

[alpinestart]

Hi -
I am faced with a situation where a particular group in my company needs/wants data within the warehouse, not cleaned, but actually changed.
They need the dates of particular sales changed so that they fall into the next fiscal period for reporting.
For whatever reason, this particular subset of sales cannot count towards current fiscal period.

I feel uncomfortable changing data to reflect a completely different reality than the source system. Should I?
I'm concerned about ever building other fact tables, having to repull, etc and things a) not conforming and b) not tying out and creating a mess, on top of
altering data to be factually incorrect.
Is this commonly done and what are other ways around it?

thanks - Rich

 

[johnherman]

You are quite justified in feeling uncomfortable in not wanting to change the real data. I suggest a separate table or tables for the groups who want to look a the data with different colored glasses than the true data which the rest of the company uses.

But seriously, these groups may have a compelling reason for having their data appear differently, so to accomodate them, we place the data in separate tables for them and use their rules for data processing (like a departmental data mart).

-------------------------
The trouble with doing something right the first time is that nobody appreciates how difficult it was - Steven Wright

 

[FacultyTDWIDAMA]

Johnherman is correct and data that is not part of the source system would be best kept separate. However, a separate database may or may not be required - you may consider just adding a field to the table that represents the data range that it should be reported on. Another, more powerful solution would be to use database recursion to allow for the changing of the data ranges on the fly by building a date hierarchy within the recursive table.

 

[cumbers5306]

if they are going to change the data so that it falls in a different time period. how are things that are built - aggregation tables, and other tables built upon the atomic data , to be modified. well, you have to give thoughts for those as well. if you create separate tables, for different set of users, then we will end up building stove pipe data marts, which cannot fit into the bigger picture for Enterprise data warehouse.EDW.