Delete tasks work similarly to IMPORT tasks, with some differences:
- Deleting based on an equality UPI value is not permitted
- No acquisition phase, because there are no variable input records to apply
- The application phase reads each target block and deletes qualifying rows
MultiLoad DELETE versus SQL DELETE:
- MultiLoad DELETE is faster and uses less disk space and I/O (no transient journal).
- MultiLoad DELETE is restartable. If SQL DELETE is aborted, transient journal applies, and rows are rolled back to the original state. SQL DELETE can be resubmitted, but starts from the beginning.
- Good for tables with a large number of rows and which are not partitioned.