For example, we are effecting major changes to one of our price planning processes. The process is roughly as follows (there are many more steps but this is just a quick overview):
- Load data from CO-PA
- Retro-engineer pricing conditions and compare to actual pricing conditions,
- Calculate aggregated prices based on complex business rules,
- Some manual planning steps
- Copy aggregated prices to detail using more complex business rules,
- Calculate new price net of rebates
- Valuate planned sales volumes using a large amount of pricing procedures.
The changes made to the data models has an impact on each of these steps and each of the steps have several planning functions containing FOX, SQL, ABAP that are impacted. Due to data volumes and the complexity of the process, errors in step 5 may only be detected at step 12 and so on, which means we have to revert back to the final state of step 4 for testing purposes.
After each planning step, we 'switch' the cube from plan mode to load mode
and back, writing the open request to the cube.
This discrete package of data now represents all the data records written in that step.
Deleting the subsequent steps' requests will put us back to the desired state. No need to re-perform all the steps from step 1 to 4.
For more tips and training on BPC, use this coupon for training on BPC