Question:
Hello BW fans
This question is related with PSA. I have an ODS which has same structure as PSA (with the same key fields). I loaded the data from PSA to ODS. Here i have a problem.
For a PO, i have 5 entries (3 entries with ROCANCEL field ' ' and 2 entries with ROCANCEL field 'X'.) But my ODS contains only 3 entries with STORNO = ' '. I need to load the other 2 entries also in my ODS.
Any suggestion is appreciated.
Thanks in advance
sampath
Answer:
In your transfer rules, is 0RECORDMODE mapped to 0STORNO or 0ROCANCEL?
_________________
Keep smiling
Answer:
Hello SR
Thanks for your reply.
0RECORDMODE is mapped to 0STORNO.
sampath
Answer:
Which ODS table are you looking at? Active table or the change log? The active table may have just the resultant rows, but the change log should have all the rows. As long as the net result of the rows are okay in the active table, then youare fine.
Answer:
Hello Guest
In both table i have only 3 entries. I need to load other 2 entries into my ODS with 0STORNO = 'X'.
Thanks
sampath
Answer:
You do not need the other entries.
The entries with ROCANCEL = X are records only to enable delta in your data target. It is not really a cancellation (STORNO in German language), but a before-image.
You should map
ROCANCEL to 0RECORDMODE and
STORNO to 0STORNO.
STORNO will then only contain S (blocked record).
ROCANCEL = R will delete the record from your receiving ODS.
Cheers
Frank
Answer:
Hello geierf
Thanks for your reply. I have the following scenario.
I want to load all the datas from PSA into my ODS including the cancelled records. Mapping is as follows :
ROCANCEL --> STORNO
STORNO --> STORNO
0RECORDMODE --> STORNO.
I am getting all records from PSA except the records which are marked with 'X' in STORNO.
Any suggestion is appreciated.
thanks
sampath
Answer:
Hi,
If any records are marked with 'X' in the 0Recordmode column, they will be skipped while loading into ODS. I think its the same case with STORNO column too when loading into ODS.
Thanks
R
Answer:
Hi sampath,
You wrote:
ROCANCEL --> STORNO
STORNO --> STORNO
0RECORDMODE --> STORNO.
Are you mapping anything to LOEKZ? This is the deletion indicator.
ROCANCEL = 'X' does not indicate a deleted record. It is a before-image and has no business background. You will get this record for every change in the logistics extractors.
To make sure that you receive all record, even the deleted, you can map LOEKZ to 0STORNO and leave 0RECORDMODE blank.
This might however screw up your counters like number of items, etc (at least this is my experience).
Cheers
Frank
EDIT:
Just to be sure: You do not want to have the whole PSA in your ODS, right?
Answer:
sampat24
i am curious to know why do you have the need to have an ODS to look exactly like a PSA.
What is the requirement ? and what are you trying to achieve ?
maybe then we could give you an alternate solution
Answer:
Thanks geierf and not_yet_a_bw_guru.
Answer for geiefr : I do not have LOEKZ in my currenct setup. In order to include this field i have to re-init again. But this is not possible right now.
Answer for not_yet_a_bw_guru : Currently, i am updating the data from PSA to ODS and then to infocube. I have several requests in PSA. During re-init, i have to update the data from PSA (one by one) to ODS then infocube. Instead of updating several requests, my idea is to load all the requests to ODS (similiar to PSA). From this ODS i can update my original ODS (using one request) during re-init. This is the motivation for this work.
Thanks
sampath
Answer:
1. are you on 2.1c or 3.x
2. what is your original data source. (to ods)
3. what is the volume of data you are dealing with ?? no. of records during your re-init.
Answer:
Hello friend here is the details
- I am in 3.1C
- Source is from R/3 system using standard extractors 2LIS_02_ITEM
and 2LIS_02_SCL.
- Currently 5 million records. In future it will increase.
thanks
sampath
Answer:
if you are in 3.1 c you can run psa in parallel but activate all at once.
you can also run oli3bw in r/3 in parallel (split it by doc number)
take care that you dont wipe out all backgroung processes.
Consider you split u'r load in to 8 (no. range for po doc) smaller loads.
you could potentially start u'r load of range 1 to bw while statistical update is still running in r/3 for 2-8.
Doing this u'r overlapping the loads and you save a bunch of time.
No need to make ODS to look like PSA ( Thats just a bad idea - i would try to Keep it simple)
5 million records is really not too much.
i have loaded @ 18 million records over a weekend.
Please also modify u'r packet size for optimal performance.
I have no idea about the size of your box or the update rules you have written - if it is a smaller box it will take more time.
Answer:
Thanks not_yet_a_bw_guru. Currently i am doing what you mentioned in your reply. But my client wants to implement an ODS with the same structure as PSA (with all records as in PSA). This is the reason i am keep on posting this.
thanks
sampath