Partition based on join in data warehouse env.

Question

Hi,
I am working on DW environment and am quite new to it.
The scenario is like there are 2 fact tables and one dimension table.
Now I need to partition both the fact tables based on the dates in Dim table. My problem is there is no date mentioned in fact table other then Surogate keys in fact tables.
Is it possible to partition the table by using join condition between fact table and Dim table and if yes how?
The structure of tables are as follows.
Please help!
Daily sales fact table
planned shipment date sk     NUMBER     PK1, FK1
source system identifier     VARCHAR2(50)     PK2
sales type indicators sk     NUMBER     PK3, FK3
ship-to-customer sk     NUMBER     PK4, FK4
product sk     NUMBER     PK5, FK5
ship-from-location sk     NUMBER     PK6, FK6
quantity in primary units     NUMBER(14,3)     
quantity in 9LE     NUMBER(14,3)     
Daily Sales Order Fact
sales invoiced date sk     NUMBER     PK1, FK1
source system identifier     VARCHAR2(50)     PK2
sales type indicators sk     NUMBER     PK3, FK3
ship-to-customer sk     NUMBER     PK4, FK4
product sk     NUMBER     PK5, FK5
ship-from-location sk     NUMBER     PK6, FK6
quantity in primary units     NUMBER(14,3)     
quantity in 9LE     NUMBER(14,3)     
Sales order month Dim
sales order month sk     NUMBER     PK1
sales order month full name     VARCHAR2(50)     
sales order month number     NUMBER(2)     
sales order month calendar year     NUMBER(4)     
sales order month financial year     NUMBER(4)     
sales order month start date     DATE     AK1
sales order month end date     DATE     
sales order month end date sk     NUMBER     
days in sales order month     NUMBER(2)     
event number     NUMBER(12)     
last update date     DATE     
Thanks in advance.

Answer 1

If you take care that the synthetic key values for you date are assigned in ascending order with the date value then you can range equi-partition by range on both of them without too much trouble.
Personally I don't use synthetic values as PK's on dates, partly for this very reason.