APD - Numbers of Jobs used to execute datasource query

Question

Hello experts,
I would like to ask for your view about an issue that I discovered while executing APD flows in background ("schedule job" button in transaction rsanwb or via Process Chain) vs. executing APD flows as dialogue job ("execute" button in transaction rsanwb).
a) If an APD flow is executed in background the datasource query of the data flow is accessing all InfoCubes (connected to the multiprovider of the query) sequentially. Only one job is used that is working its way through around 10 InfoCubes after each other.
b) If an APD flow is executed in background the datasource query of the data flow is accessing all InfoCubes connected to the multiprovider in parallel using many jobs that are reading the content of the InfoCubes at the same time.
It would be much appreciated if you could provide some input to me how this behaviour is driven and how it can be influenced. From performance point of view I would like the query in the APD Flow to read several InfoCubes in parallel instead of processing each InfoCube after the other.
Thanks for your help.
Cheers,
Sven

Answer 1

Hi all,
in the following SAP note "Note 605213 - RSCRM: Performance" it says that ODBO (when an APD Analysis Process is executed, the required data is read via the ODBO interface of SAP BW) executes the query in a different way than BEx or TA RSRT does. So you can't compare both regarding speed.
Still I am wondering why there are difference if I execute the Analysis Process in background job and dialogue in the way the query is accessing the InfoCubes of the used Multiprovider in paralle (dialogue) or sequentially (background).
Any ideas? You help is much appreciated.
Cheers,
Sven