A real challenged task in BW - Only experts can solve it

Question

Hi BW experts,
I get a new task, to create a cube for analyzing our products at the customers.
The customers send us some files that contain some products details.
For examples, let's say thta the products are computers and the files are:
1. Operation system
customer ID, product ID, OS name, OS version
2. Firewall properties
customer ID, product ID, FW property name, FW property value
3. Browser details
customer ID, product ID, browser name, browser version
Each customer ID and product ID may have more than one record. In other words they not the key of the table.
The questions that we would like to ask, are:
1. How many many customers have OS name = Linux.
2. Which customer have product ID = 111, and fw_property_name = zzz and browser name = Mozilla.
3. How many customers have OS_name = windows AND
browser_name = mozilla.
Can you please assist of it?
Best regards,
Mohamad.

Answer 1

Hi Philipp,
Thank you for the replying.
Please look at the example below;
1. Operation system
customer ID, product ID, OS name, OS version
111 , comp01 , Window , XP
2. Firewall properties
customer ID, product ID, FW property name, FW property value
111 , comp01 , FW_Protection_extent , Full
111 , comp01, Is_AntiSpy_switched_on, true
3. Browser details
customer ID, product ID, browser name, browser version
111 , comp01 , Internet Explorer , 7.0
111 , comp01 , Mozilla , 2.0
As your suggestion, the fact table should be like this:
customer ID, product ID, OS name, OS version, FW property name, FW property value, browser name, browser version , KeyFigure(Fix value = 1)
Therefore, we need to perform a JOIN between all the files above, and then load them to the cube. What we will get is;
111,comp01,Window,XP,FW_Protection_extent,Full,Internet Explorer , 7.0 , 1
111,comp01,Window,XP,FW_Protection_extent,Full,Mozilla , 2.0 , 1
111,comp01,Window,XP,Is_AntiSpy_switched_on, true,Internet Explorer , 7.0 , 1
111,comp01,Window,XP,Is_AntiSpy_switched_on, true,Mozilla , 2.0 , 1
This will solve our issue, but the problem of this solution that if each file contains 1000 records, then the JOIN will give 100010001000 records. Therefore, it's not legitimate solution.
Best regards,
Mohamad