Tipe 2 Dimension 1

[ribula]

Hi!
Vi have a long disskation obout stategi to bild dimention.
Most of all dimentions vi have are vary big...
(Big Bank with allot customers).
Vi allrady resolve to bild Type 1 when dimensions is
5-10 mil rows
Should vi have type2(slow change di on small one?
When Type2 is god to use or not. Which reason to use them?
Thanks

 

[johnherman]

Type 2 dimension is important when the customer has a need for the prior values of the dimensional attributes.

For instance, if a customer changes the phone number, does the bank care? NO, the bank will not be calling the old phone number.

If a customer changes their name (say for marriage), does the bank care? YES.

If a customer changes their address, does the bank care about the old address? MAYBE - only the bank can answer this question.

Maybe and Sometimes is the best answer to your question about when to use Type 2 Slowly Changing Dimensions. The users of the data warehouse (bank people) must tell you what changes are important to track through time and when previous values do not matter and do not need to be saved.

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

 

[ribula]

Thanks vary match for explain!
Mi expiriens with DW during 10 y
The problem is, when I make choise use Tipe 2 or when is better use FAKT for make history.
As usaly vi have vary big dimensions (castomer,Agrement, product_agriment, department e.t.c, between 10 mil up to 30 mil rows)
It's not god for analysis to scan big dimension, much indexis dasn't help, DW on Z/OS DB2 V8 wich not so mach to Starshema.Thats why my questien was if you have expiriens to make 'history' on Fakt. Which for or not for this aproch
Best Regards

 

[johnherman]

Always remember to design you data warehouse based on business requirements. If your business requirements show a need to query large tables rapidly, you might consider having a dimension history table and a dimension (current) for each large dimension. That would be ideal if the DW users are mostly interested in the current view.


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