The most common use of row triggers is with OLD and NEW correlation names that reference current row values as scalar quantities. Row triggers provide the following advantages:
- The restriction that each row is processed only once guarantees that trigger executions are independent of one another.
- It is not generally necessary to wait for completion of the triggered action for a given row to initiate action for the next row.
- Because OLD and NEW correlation values can be accessed prior to dispatching the triggered action steps, most cases involve sending single AMP steps, so they do not require full table locks.
If the triggered action for a row changes data values in such a way that the WHEN condition search results for some rows might be affected, then the system must wait for completion of the triggered action before proceeding to the next row. A trigger with such characteristics cannot take full advantage of the Teradata parallel architecture.