Generally, we use 2 DSO's when we have a large data volume, for ex. logistics.
The 1st DSO is the exact replica of the Source system data, normally a Write-optimized DSO, where-in the data is dumped without much/ no transformations. Also, there is no need of activating the DSO.
In the 2nd level DSO, the transformations/ rules come into the picture and it is more of the required data/ fields and not the entire data set. Like this we ensure that the entire source system data sits in the 1st level and the required data is loaded into the further levels. The 2nd level DSO's are generally Standard in nature.
Coming to your question of should we always use 2 level staging - i would say that really depends on what you're doing and what the exact requirement is. For ex. you can have the header and the item details in different DSO's in the 1st level and then consolidate both in the 2nd level DSO before loading the data into the cubes.
Please let us know what the exact requirement is.
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.