Extraction from a cube slower than extraction from an DSO
maybe a strange question but I will ask it anyway
We got a pure data integration scenario where we will load data to another BW system.
For this we build an Outbound Infoprovider that has the format as the other BW system needs it - all the mapping tables for transformations are residing in the "sending BW". On this Outbound infoprovider no reporting will happen at all.
As the flow is very time critical the question arised what would be better in terms of extraction performance a cube or a DSO.
If we run an extract on an cube DIM tables would need to be joined together to produce the result.
On an DSO it would be a pick the record and sent.
Therefore is in a pure data integration scenario the DSO the better choice from an performance perspective or won't we see any difference performance whise cube vs. DSO extract.
Edited by: Axel on May 18, 2011 10:23 AM
To add to it, if your source BW system is just a pass through then it would be even better if you use a Write Optimized DSO to stage the data and then send it through outbound interface, as you will save time in activation process.