Do you know, if Load from Data stream automatically should be able read an aggregate, if it finds an aggregate that has the required fields - or do you know if we can force it to read the aggregate by initiate certain actions?
If this is not possible, then we probably have to make a staging cube, which physically holds the same data as aggregate and then use Load from DataStream on the staging cube and this way force it to read fewer data.
Some background, which you may skip or may read (as the question should hopefully be clear enough): At one big project we have been live with R/3, BW, BCS and BPS system since 2006, there are some of the big company codes recently being rolled out, where Load from Data Stream takes very long time (and often dumps with time out).
In the beginning some years ago, we seemed to have similar problems when using Load from Datastream from BPS to BCS - but when we then made aggregates on the BPS cube it seemed to help. Even though we were never able to prove it technically, then making the aggregates seemed to make the run times much faster, but maybe it was a coincidence.
Now when are getting problems on our actual data being read with Load from DataStream to BCS from our Profit Center Accounting cube, we tried to make aggregates on the PCA cube - but this time we did not seem to improve performance.
I could not find anything in the documentation about using Load from DataStream on aggregates and did not find any notes about it in OSS and no info in SDN either (eg when searching on Load from datastream and performance or aggregates).
Thanks and best regards Jesper Schou Nielsen
Edited by: Jesper Schou Nielsen on Dec 4, 2009 9:07 AM