2.0 Storage of non-observation attributes at series and observation level
Change the way attributes are stored in the data database. Currently the attributes are stored in different tables depending on their attatchment level:
- Observation level attributes are stored in the fact tables
- The rest are stored in the [data].[ATTR_{DSD ID}_{Table version(A|B)}_DIMGROUP] table.
The task is to keep observation level attributes in the fact tables, and the rest store them in in a similar manner as a flat representation.
-
Create new table [data].[ATTR_{DSD ID}_{Table version(A|B)}] - Primary key [SID] from filter table
- One column per attribute with the name pattern [COMP_{incremental number (1-n)}]
-
Change the import process not to import attributes in a separate bulk insert - All dimensions, measure and all attributes should be copied directly into the staging table
-
Adjust DSD and Dataflow views to adopt new structure -
Include SQL scripts to migrate existing users -
Update the database model documentation
Background documents on analysis done for .Stat V7:
Edited by Jens Dossé