Skip to Content
0
Former Member
Nov 26, 2006 at 09:50 PM

Design of data-flow

33 Views

Hello experts!

We want to design our data-flow in a new way and I want to know from some experts, if this is a good way of if there maybe is a better way.

At the moment we extract over the following InfoSources:

0CO_OM_WBS_1

0CO_OM_WBS_6

0CO_OM_OPA_1

0CO_OM_OPA_6

We extract from every data source in full load in up to 8 different InfoCubes. Everytime we update different attributes and selections with a lot of self written ABAP-rules and so on. All historical grown.

Now we don't want to read the data every time from DataSource into BW. So we say, we need an entry layer end update all the data from inside the BW.

We know choose an InfoCube as update-layer, because we maybe want to report on this new data befor it is update to the data targets. We also want to delete requests which could be difficult with an ODS-Object. An ODS-Object has also just 16 key-fields which could be a problem.

So know we want to design an InfoCube for every InfoSource and make data mart update to the other InfoCubes ( data targets).

What we also want is a additional layer for saving some requests sometimes. At the moment I would say, I just make a copy of the InfoCubes I want to design.

So thats it. For design of the InfoCubes I have an additional thread:

InfoCube Design

I have just around one year of BW-experience and never done this before. Maybe someone can give me some tipps or hints if this is a good way or if there is a better one.

Thanks in advance and best regards,

Peter