A prospect insists that pulling data from the SSAS Cubes is the best idea rather than building the data model inside QlikView and replicating the transformations they applied. Their reasoning is that building the data model inside QlikView will create another validation headache. They also want to keep it as people don't want to stop using Excel PivotTables. I very much understand this. Given the situation, this is what I have proposed.
1. Export a snapshot of the Fact Table to CSV and Export to QVD.
2. Export a snapshot of the Dimension Tables to CSV and Export to QVD.
3. Setup Incremental Load for the Fact Table using MDX (read from Cube). This is done manually as QV cannot create MDX queries.
4. Setup Incremental Load for the Dimension Tables using SQL (read from DB).
Problems I foresee:
1. When the cube is rebuilt, the QV application has to be remapped (map new fields to existing fields).
2. In the wors case scenario where the cube does not resemble the previous cube, the QV application has to be built again. Not a lot of brain work, but more on leg work.
What other problems do you guys foresee given the situation? Any alternatives to propose?
- Kevin
. Thank you...