Hi Cheeli,
in the Aqcuisition phase the rows are loaded (similar to a FastLoad) into the worktable, one row for each APPLY WHERE.
At the end of this phase the worktable is sorted by the so-called "match tag" columns (automatically added) which ensures that all modifications are done in the correct order based on input sequence number.
Both target and work table have the same PI and during the application phase all those modifications are merged into the target using a sequential scan on both tables.
So your friend was not really right, the data for the inserts/updates/deletes are recorded in the worktable, but they're actually done on the target rows.
If you read the MLoad manual there shoud be more detailed info.
Dieter
Hi Cheeli,
in the Aqcuisition phase the rows are loaded (similar to a FastLoad) into the worktable, one row for each APPLY WHERE.
At the end of this phase the worktable is sorted by the so-called "match tag" columns (automatically added) which ensures that all modifications are done in the correct order based on input sequence number.
Both target and work table have the same PI and during the application phase all those modifications are merged into the target using a sequential scan on both tables.
So your friend was not really right, the data for the inserts/updates/deletes are recorded in the worktable, but they're actually done on the target rows.
If you read the MLoad manual there shoud be more detailed info.
Dieter