We have a Tabular Model that exists in Azure - processed multiple times a day to catch intra-day transactions.
For simplicity sake I will denote the Table Structure as:
FactTable1 (Order Transactional)
FactTable2 (Detail Transactional)
DimTable1
DimTable2
DimTable3
DimTable4
DimTable5
FactTable1 and 2 are getting rather large and so I wanted to look into Processing by Partition - to being we can have enough of a gain by Partitioning by year (which I know how to do). The question now comes to Processing the Tabular Model - If I have SCD in the Dimension Tables or the addition of new rows - how do I ensure that is processed along with the Fact Tables.
Currently we use PowerShell/Webhooks to process on a certain cadence -
Invoke-ProcessASDatabase -server "asazure://servername" -DatabaseName "DatabaseName" -RefreshType Full
First question: If I wanted to process FactTable1_2021/FactTable2_2021 daily - do I split the command into one command for the individual tables and then the rest as a RefreshType Default? How do Dimension Tables (1 - 5) get processed?
Second Question: If I wanted to go back and process FactTable1_2020/FactTable2_2020 weekly (there is a small chance of historical changes) - can I process that between my daily processes without any issue?
I have not seen any clear diagrams of how Dimensions process - is it just when a record in a FactTable shows up that uses the relationship ship it checks for an existing record?
Thanks for any help,
Derek