Configure a target connection to connect to your target. Configure the target connection on the Connect Target page. You can create a new connection or select an existing connection.
The following image shows the Connect Target page:
To create a target connection, click New Connection. You can also select an existing connection. After you choose the connection, you might have to configure additional properties like Write Disposition and Staging Location that vary based on the connection type. For more information about connection properties, see Connections.
The first time you run a data loader task, the task creates one target table for each source object. By default, target table names are the same as the source object names. However, special characters in the source table names will be replaced with underscore characters. For example, if the source table name is Orders$, the corresponding target table will be named Orders_.
If you enter a target name prefix, the target object names will be the same as the source object names, but they'll be preceded with the prefix. For example, if a source table name is Account, and you enter the prefix tgt_, the target table name will be tgt_Account.
When you re-run a task, you can either load data to the target incrementally or drop and re-create the target tables.
Incremental loading
You can configure a data loader task to load data to the target incrementally. When the task loads data incrementally, only new and changed data is loaded to the target each time you re-run the task. Incremental loading increases task performance since fewer rows are loaded.
To configure the task to load data incrementally, select Yes under Load to existing tables on the Connect Target page.
When you load data incrementally, the task runs most efficiently when you configure the following options in the source:
Primary key fields are detected automatically or entered manually.
The task uses the primary key fields to identify rows when updating or inserting data into the target tables. If there are no primary key fields, the task can insert data into the target tables, but it can't update existing rows. This can create duplicate rows in the target tables.
Watermark fields are detected automatically or entered manually.
To load data incrementally, the source objects must contain watermark fields. The task uses the watermark fields to determine which records have been added or changed since the last task run. If there are no watermark fields, the task can't determine which records are new or changed. So, it processes all records in the source objects each time the task runs.
If the source object structure changes between task runs, the data loader task detects the changes and alters the target tables to match the source objects. Therefore, if you add a field to a source object or change a field's data type, the task adds the new column or updates the column data type in the corresponding target table. The task does not delete any existing target fields, however. If you want to delete unnecessary target fields, you can either delete the fields manually or configure the task to drop and re-create the target tables.
If the name or data type of the watermark column changes between task runs, or if you change the watermark column between task runs, the task performs a full load even if you've configured the target to load to existing tables.
Dropping and re-creating target tables
You can configure a data loader task to drop the target tables and re-create them every time the task runs. When you select this option, the task deletes the target tables created by the original task run and creates new tables using the schema of the source objects.
To configure the task to drop and re-create the target tables, select No, create new tables every time under Load to existing tables on the Connect Target page. To ensure that the task processes all records each time it runs, you also need to select Watermark field not required under Define Watermark Fields on the Connect Source page.
When you select these options, the task performs a full target load each time it runs.
