Re: Help with DB_LINK parallel insert as select

Billy Verreynne wrote:

>
> Try the following:
> insert /*+ append parallel(dst,4) */ into working_raw_marks_test dst
> select
> /*+ parallel(src,4) */
> *
> from working_raw_marks_at_msup_link src,
> rowid_table rid
> where src.rowid = rid.row_id

Hmm.. I missed that dblink in the SQL. That changes the picture somewhat. Distributed joins is a thing that I never do - I have more horror stories than good stories about it. So I very very seldom consider doing a distributed join.

When dealing with data processing that includes grabbing data via dblinks, I usually just copy the data across with a CREATE TABLE NOLOGGING AS SELECT.

In that case, the above SQL should work okay with as the raw marks table is local.

The dblink also explains why the SQL slows down when adding the WHERE clause and does not parallise.

--
Billy