Loading Process and Loading Table Structure in odi

Loading Process

A loading process is required when source data needs to be loaded into the staging area. This loading is needed when some transformation take place in the staging area and the source schema is not located in the same server as the staging area. The staging area is the target of the loading phase.

Loading Process Overview

A typical loading process works in the following way:

1. A temporary loading table is dropped (if it exists) and then created in the staging area
   
2. Data is loaded from the source into this loading table using a loading method.
   Action 1 and 2 are repeated for all the source data that needs to be moved to the staging area.
   The data is used in the integration phase to load the integration table.
   
3. After the integration phase, before the interface completes, the temporary loading table is dropped.
   
   
   
   

   Loading Table Structure

   The loading process creates in the staging area a loading table. This loading table is typically prefixed with a C$.
   A loading table represent a source set and not a source datastore. There is no direct mapping between the sources datastore and the loading table. Source sets appear in the flow tab of the interface editor.
   The following cases illustrate the notion of source set:
   
4. If a source CUSTOMER table has only 2 column CUST_NAME, CUST_ID used in mapping and joins on the staging area, then the loading table will only contain an image of these two columns. Columns not needed for the rest of the integration flow not appear the loading table.
   
5. If a is CUSTOMER table is filtered on CUST_AGE on the source, and CUST_AGE is not used afterwards, then the loading table will not include CUST_AGE. The loading process will process the filter in the source data server, and the loading table will contain the filtered records.
   
6. If two table CUSTOMER and SALES_REPS are combined using a join on the source and the resulting source set is used in transformations taking place in the staging area, the loading table will contain the combined columns from CUSTOMER and SALES_REPS.
   
7. If all the columns of a source datastore are mapped and this datastore is not joined on the source, then the source set is the whole source datastore. In that case, the loading table is the exact image of the source datastore. This is the case for source technologies with no transformation capabilities such as File.
   
8.