Potential Impacts of Configuration Changes to the Data Warehouse in existing Systems

note_awesome

NOTE

The information below is only relevant for 'extra-ordinary' changes to the configuration, such as changes to

Component Names
Field and Relationship Names
Cardinalities

'Regular' changes to a configuration will be handled without any of the following implications, or required steps.

Label Changes to Components, Fields and Relationships

WARNING:

Changing the name of a Component, and then performing the steps below, will have the consequence that reports using the CUBE for this component will need to be deleted, and recreated from scratch.

Alternatively, the name change needs to be reverted.

The initial steps to propagate a label change are identical, regardless of the field type:

Perform your changes in the Component Management as required.
Run a Publish.
Run a Publish BI.
The following step depends on the field type which was renamed:

Field Type(s)	Is renaming of old column name to new column name in the data warehouse required?
Auto Number Text Date Time Integer Decimal Single Lookup List Matrix	Update the following tables: `ft_[component name]` `se_[component name]`
Memo	Update the following table `se_[component name]`
Multi Lookup List	New tables are created automatically. You may choose to clean up old tables and views which are no longer used.
Single Relationship	Update the following table `ft_[component name]` The system will create new `lk_` and `ls_` tables, but you may choose to clean up old tables which are no longer used.
Many to Many Relationship	New tables are created automatically. You may choose to clean up old tables and views which are no longer used.

Generic SQL scripts for renaming and cleaning up of database tables can be found under this link.
ONLY if the component involved is a Rollup Component:
Drop the components dimension tables. The ETL will rebuild and repopulate the tables. This is needed to ensure tables are populated correctly and to ensure the old legacy columns are removed.
Run the ETL Process
Import the Cube
If any of the altered fields/relationships is used in a report, then it will be necessary to update the SQL used in the report, and/or to update the report which is using the Cube.

Changes to Cardinality of Relationships

Cardinality Change

Impact on Data Warehouse

Changes from

1:1 to 1:many
1:1 to many:1
1:many to 1:1

As long as the MANY side is the Primary side of the relationship, then the data will be retained in SAI360 and therefore the Data Warehouse.

If not, the data in SAI360 will be lost, new columns will be created in the Data Warehouse and therefore will require remapping.

The old columns will need to be cleaned up.

Changes from

1:many to many:many
many:1 to many:many

Data will be lost in SAI360, but the Data Warehouse will have a new view that still holds the old data, as it references its link and link satellite tables.

The user may choose to clean up old tables.