Best approach for Data Modelling.

Question

Hello Experts
I am building a Customer Scorecard involving SD and Marketing in BI 7.0.
There are a couple of existing DSOs, some pushing the data into InfoCubes and some don't. All the reporting is happening from MultiProvider sitting on top of these Data Targets.
The team has a primitive design which says that I additional DSOs be created to extract data from the above mentioned couple of DSOs based on only the Objects that are needed for Customer Scorecard reporting.
This means, I am creating a couple of DSOs as per the current design which is in place.
Upon suggesting to only create a Customer Scorecard MultiProvider on top of the already existing couple of Data Targets (avoiding to recreate addtional DSOs and the hassles of loading and activating them and then loading the data into InfoCubes) and then create the BEx Queries on top of them, the Lead expressed his concerns about the impacts it could have on the existing Data Model and subsequent transports once the Model is complete..!
What is the best practice to handle a situation like this? I see there are 3 ways to go ahead with this:
1. Do as the Lead said, which means creating additional DSOs (extracting data from a couple of required existing DSOs, push this data into 1 InfoCube and then create a MultiProvider on top of this (be aware that there is another similar data model that I need to create which will also be embedded into this MultiProvider) and create BEx Reports from there.
2. Create only the InfoCubes which will extract data from the already existing DSOs (avoid creation of additional DSOs) and then create a MP from where BEx Reports are created.
3. Only create a MultiProvider on all the required and already existing DSOs and InfoCubes, making sure if reporting needs aggregated data for reporting or not and then create BEx Reports from there (avoid creation of additional DSOs, & ICs).
Note: We use Rev-Track to do the Transports.
Which one do you think would be the best way to go and what could be the implications? Eventually, the reporting is done in WAD.
Thanks for your time in advance.
Cheers,
Chandu

Answer 1

Hi,
Case 1 and 2 have similarities. But its purely depend user needs.
I think you may be know the difference between dso and cube.
DSO - holds detailed level data
Cube - holds aggregated data.
As per you needs use any one target only, no need to use DSO---> cube flow for existing flows.
you can decide which you want use DSO or Cube only.
Case 3. if your requirement will suffice with existing dso and at reporting level if you can manage to get the required out put then you can with it. But as my guess with existing target your requirement may won't suffice your needs.
About transports:
You can create one Rev track and assign multiple transports to it.
you can add and release transport one by one rather than all at a time.
if you release all at a time you may get some inconsistency issue and TR won't be released.
Thanks