Multiple fact tables, dealing with snowflake data models

[Koen_D]

Hi Qlik experts,

We have a lot of data coming from multiple sources that need to be linked. Each source has its own data warehouse that we load into Qlik. From each source we load data that already has a snowflake scheme. We then have to combine these tables in Qlik into one big data model, multiple fact tables with a lot of dimensions. 

I know Qlik works best using a star scheme datamodel. I normally concatenate my facttables with a FactType field, but in this instance I need to be able to filter all of this data on any dimension. The front end is rather slow now, so I'm looking to improve the performance of the dashboard.

My questions are:

1. What is the best way to deal with these multiple fact tables?

2. When dealing with a snowflake scheme, is it better to transform the model into a star scheme? For example, an Order table that is connected to a Warehouse table which then again is connected to a Location table, would it be better to add a location field to the Order table? Or just left join the Location table to the Warehouse table?

[marcus_sommer]

In regard to the UI performance a star-scheme data-model will perform better as any data-model with multiple fact-tables and I remember on suggestions to use a big flat-table by really huge amounts of data.

Beside this you should look to optimize the data-model and the UI design, for example:

• not to loading record-id's
• minimizing the number of distinct field-values by fields like timestamps by splitting them into dates and times
• not using a row-level formatting for numbers and dates
• not applying nested if-loops, aggr-constructs and interrecord-functions within the UI