5 Transformations
- Use the Regenerate program definitions option to rebuild program definitions in a transformation
- Remove a source model from its transformations before deleting it
- Avoid using the Undo Checkout feature for transformations
- Use unique aliases
- Do not copy a transformation that needs an upgrade
- Wait to create staging tables
Use the Regenerate program definitions option to rebuild program definitions in a transformation
-
Recommendation: We recommend that you select Regenerate program definitions when you run a regular transformation installation if you want DMW to rebuild program definitions based on the latest version of the source and target data models in the transformation.
-
Rationale: If you select Regenerate program definitions when you run a regular transformation installation, DMW does the following:
-
Rebuilds the program definitions based on the latest version of the source and target data models.
-
Regenerates portions of the program that were missing prior to the modifications.
-
-
Additional information: If you are running a full transformation installation, we recommend that you only select Regenerate program definitions if absolutely necessary.
Parent topic: Transformations
Remove a source model from its transformations before deleting it
-
Recommendation: If you need to remove a data model from a study:
-
Check out transformations that use the data model as a source model.
-
For each transformation, click Add or Remove Source Models.
-
In the Add Model dialog box, deselect the source model that you want to delete.
-
Delete the data model from the study.
-
Parent topic: Transformations
Avoid using the Undo Checkout feature for transformations
-
Recommendation: Instead of using the Undo Checkout option for a transformation, check the transformation in and then check it out again.
-
Rationale: When you use the Undo Checkout option, DMW might remove custom functions associated with the transformation. In addition, other issues related to expressions could occur.
-
Additional information: We recommend tracking all changes as new versions in DMW. You can always roll objects back to their most recent Production version.
Parent topic: Transformations
Use unique aliases
-
Recommendation: When you use the auto-map feature, we recommend that you select a single target table. We do not recommend using the auto-map feature to map at the data model level.
-
Rationale: Poor performance occurs when you auto-map more than one table at a time, and is worsened when you map to 10 or more tables at a time.
-
Additional information: To ensure the best performance when you auto-map, use the Column Mapping user interface, and use the Shift+Select keyboard shortcut to run auto-mapping only for columns in a single table that have known map matches.
Parent topic: Transformations
Do not copy a transformation that needs an upgrade
-
Recommendation: When you copy a transformation from one study to another using the Copy from Another Transformation option, make sure that the transformation you want to copy does not require an upgrade.
-
Rationale: A transformation that requires an upgrade references old data model definitions. As a result, if you copy it, you'll likely need to perform manual updates to the copy.
-
Additional information: To further reduce the likelihood that you'll need to make manual changes to the copied transformation, we recommend that you copy transformations that were created by templates or study libraries. However, we also recommend that you manually confirm that the transformation doesn't contain maps that need to be upgraded.
Parent topic: Transformations
Wait to create staging tables
-
Recommendation: Create staging tables only when the Aggregation Layer and the desired Consumer model are defined, and the transformation is known.
Avoid defining staging tables on the fly within the target data model.
-
Rationale: The Create Staging Table functionality has limitations. As a result, we do not recommend using it.
-
Additional information: The Guidance for DMW Implementation and Configuration document contains details for creating staging data models that partition the intermediate steps from the target models that are intended for consumption.
Parent topic: Transformations