I recently came across a question regarding the organization of the transformation layer of my Qlikview projects.
After I extract the data I need from our ERP, I have one application per object, where I created dimensional and facts qvds (for optimized loads in the load layer). For instance, I have:
- Customers.qvd
- Products.qvd
- SalesOrders.qvd (which creates DIM_SalesOrders.qvd and FAC_SalesReport_SalesOrders.qvd for instance)
- SalesDeliveries.qvd
- SalesInvoices.qvd
etc
And I reload these applications one after the other.
But I was recently asked to create an application on service rate, for which I need data from both sales orders and sales deliveries. Now, my first reflex would have been to create a new fact table in SalesOrders.qvd, because this service rate needs to be calculated by order line. However, to do this, I needed data from SalesDeliveries.qvd which was not updated yet, so I ended up creating my new fact table in SalesDeliveries.qvd.
So, what is the recommendation? Should I have one transform application per object, or put them together into one big application?
