Istvan Veres wrote:
Improving data error becomes more complicated as well.
Actually, it will be less complicated if you do it right... the search for the error might be more complicated, but the solution could be a lot easier. Example you partition on company code (or country), if there is an error for 1 company code only, you only need to "fix" (possibly reload) one part of the flow (instead of the entire flow).
Istvan Veres wrote:
My question is:
What amount of overhead on the system's resources (eg. BGD processes), and the operators work?
What are the pros / cons of using this?
Overhead will depend on your scenario and your system landscape. Some customers have limited background processes available, obviously if that's the case, then you may not want to implement this without thinking it through. You then have to schedule the various "subflows" in serial (instead of parallel, which I think is the default).
Pros: are definitely error resolution, reporting should be (a lot) faster (e.g. partitioning on fiscal year for COPA - most reports only access this year's and last year's data... so 2 partitions, of possibly 14 if you started with COPA in 2000), ... You could also keep data flows per continent for example (different continents have different business hours). Authorizations might be more "visible" (if you set them per flow).
Cons: you'll have more data providers, but that's not really a disadvantage as such.