Question:
Hi all,
This is really scary.
I have uploaded 20000 recrods from a flat file into a cube. I have ran a report and total of the Amont is not coming correctly. So, I went to the monitor and looked at the status and found (20000 from 20000 records). Then I wnet to manage and to fact table content and listed the number of records, it said 19998. Isn't this weird, where monitor shows that 20000 records from 20000 records transferred to the cube and the actual infocube contents are less. So, I reloaded the cube and this time it went fine. I got the totals correct between the source and the infocube reports.
Have anybody faced this isutation. Also, is there a way I can this problem. Is it because of the incorrect SIDs? Do I need to contact SAP. I do not guarantee that same thing repeats again. Also, for 20000 records I could check. What happens when millons of records get transferred and redoing th ewhole load is time consuming and it is not right.
Also, I have another case, where we have a ODS inbetween the Cube and the infosource. ODS gets loaded properly. In th eCUbe, I found that the transferred records and Added records count is different. I monitor job status tells that "n from n records".
Thanks in advance.
- Gary
Answer:
Difference in transferred / updated records can happen, e.g. if records are filtered out in the start routine, or if additional records are ceated here. Also, if the characteristic mapping is different for different key figures in the update rules.
Number of loaded / stored records can be different. Because o fthe above. Also note, that FactTabes (F tables) allow for non-unique primary keys.
Of course, the query results should be 'correct'. I'd recommend to use RSRV (and potentially aggrgate check tools) for validation.
In general, it is always a good idea to build in a automated data checks. Boing does it, Airbus does it, Space Shuttle does it - they all have excellent software, however believe in redundancy for quality assurance.