Archives of the TeradataForum
Message Posted: Thu, 21 Mar 2002 @ 13:37:09 GMT
| Subj: || || Re: RowID join with condition of (1=1) |
| From: || || Adam DeYoung |
I'm looking at SQL that does the row id join (1=1) right now. Look for a prior step accessing the table via a covered index (traversal
of index #n extracting row ids only) - it is scanning the index instead of the table to give it the rows it is looking for. The row id join
step is where Teradata actually retrieves the rows from the table by joining the row ids from the index to the row ids on the table. It's
definitely not a product join, it's a much faster row id join.
Product joins can also show as (1=1) joins if no columns are related between the two tables, but it explicitly tells you the join is a
product join in that case.