0_FI_AR_4 Modeling question

Question

Hi ,
How does one model a solution in BW for getting a correct  historical update of open and closed items from 0fi_ar_4.
Imagine that the data is staged upward from write optimized dso to a std. dso and going up the cube.
Now at the datamart level, how do we ensure a consistent updated picture of open closed items as they change from one status to another?
Thanks

Answer 1

Thanks Simon,
I like the idea about two seperate cubes for Open and Closed.
Our existing design is as follows:
the five Write optimized DSOs (WDSO) - one for each source system,  feed data to one line item Standard DSO (SDSO) . Now the KEY for this SDSO includes Open/Closed Indicator. 
The SDSO feeds(1 DTP) a Cube for AR Line items (CubeA- doesnt have source system field) and also feeds a Transformation DSO(TDSO) which aggregates into Total AR, monthly debits and current and past due on line items. and also a couple of lookups.
This TDSO feeds up into a Cube for AR monthly (CubeM).
The design team that developed this model is not reachable anymore and I am working towards understanding the workings and conveying some performance fixes for this model, which could also entail a complete redesign.
Now, they have 5 process chains, for each source system, each triggered by loads into the WDSOs.
Each PC has two parts :
1. Open item selective deletion
2. Closed item selective deletion
The explanation for the numerous selective deletes everyday is :
Everyday, there is a full load of Open items and only current year records of Closed items. So, its necessary to delete and reload the Open items and current year records for Closed items from WDSOs, SDSO and CubeA.
For monthly snapshot to be accurate, it is necessary to delete and reload current month data in TDSO and CubeM.
Now in the absence of any real knowledge transfer and at the behest of making a decent evaluation/recommendation, I need to understand why they have the current model andmultiple selective deletes.
My observations:
- The Open/Closed indicator cannot be a part of the key for SDSO as it would result in doubling up of the key figures.
- There need not be Selective Deletion for the Open Items, iff source system field is included in CubeA and source system specific DTPs are employed.
- To eliminate the Selective deletes for Closed items, 'Delete Overlapping Requests' (DOR) type can be used in Process Chains. To account for building history one month at a time (rolling 12 months), the DOR could be coded in the routine.
Please let me know based on the above description, if my observations are correct. And what would it cost to make the suggested fixes.
Also, there should be a better way of building this model. Please let me know your thoughts. Appreciate your insightful replies!