The delete task uses a full file scan to remove a large number of rows from a single database table. (A delete task cannot be used on a view.)
If a database restart/recovery operation interrupts a Teradata MultiLoad delete task, Teradata MultiLoad resumes processing from the point of interruption. It does not roll back and repeat processing that took place before the interruption. This effectively eliminates the processing penalty that would be incurred by transaction rollback during a restart/recovery operation.
The following table describes the considerations when programming delete tasks in the Teradata MultiLoad job script.
| Topic | Delete Task Considerations |
|---|---|
| Archive/Recovery Utility Statements | These statements include:
Teradata MultiLoad treats these the same during the application phase of a delete task as during the application phase of an import task, as described in the preceding topic. |
| Index Specifications | A delete task can specify a unique primary index (UPI), a non-unique primary index (NUPI), or a non-unique secondary index (NUSI). A unique secondary index (USI) is not allowed. |
| Locks | The locks protocol for a Teradata MultiLoad delete task is the same as for an import task, as described in the preceding topic. |