Transformation Processes (DMM)

A transformation process is an instance of a data transformation engine that extracts data from input sources, transforms it, and loads it to output sources. Input and output sources can be databases, flat files, XML documents, or business processes.

Note: You should use a transformation process when your main focus is complex transformations of data, such as are required for data warehousing. For simple copying of data, you should use a replication process (see Replication Processes (DMM)).

In the following example, multiple input sources are transformed by the Data Fusion and Reorganization transformation process, and then loaded to the Giant Corp data warehouse:


transformationprocessschema.gif

Note: You must deploy your transformation process to a server (see Servers (DMM)) to ensure correct script generation.

Although you can create all the objects necessary to model a data transformation by hand in any order, we recommend that you use the following workflow:

  1. Identify any input and output sources. These may be existing PDMs, XSMs, BPMs and flat files or live data sources that can be reversed engineered.

  2. Create a DMM and launch one of the following wizards to create your basic transformation environment:
  3. Press Ctrl, and double-click the transformation process symbol to open its created data transformation diagram, and specify any other necessary transformation objects, such as data query executions, calculators, etc. (see Data Transformation Diagrams).

  4. [optional] Create a control flow diagram to detail the order in which a series of data transformation tasks is executed (see Transformation Control Flow Diagrams).

You can use the completed model to guide you in writing a data transformation for your ETL engine.