Checking for duplicate primary keys on row inserts

Question

Checking for duplicate primary keys on row inserts
I have a situation where I will be making bulk table inserts knowing that the primary key value will in some cases already exist. In this is the case I simply want to ignore the duplicate inserts.
Should I be performing a sub-query on the table and using a statement like:
where not exist in
Or is there a cleaner way of discarding or checking for duplicates on insert.
My concerns were mainly one of performance, as my routine will be inserting a few thousand rows in its operation.

Answer 1

The MERGE commnad is a good option when a large percentage of the data will exist in the target because it is much more efficient to attempt to update then insert when the update affects zero rows than capture an error and convert it to an update.
However, since in this case it would appear that only a few rows will alreadys exist and we want to ignore the duplicates when they exist then
begin
insert
exception
when dup_value_in_index then null;
end
would be the way to code this one. The bulk insert version has in 9.2 the ability to store the errors so that they can all be handled at once which means the rest of the array insert can work.
HTH -- Mark D Powell --