Re: Table Joining Sequence

Yeah, Bakker, you are right. It seems that Norris is using the rulebased  optimizer. Nowdays, most of the mainstream RDBMSs have switched to cost-based optimization.
As for "join" operation, most optimizers not only re-order the join sequences, choose the best join-predicates, but also choose best join algorithms. So createing index for some columns will speed the join, because the optimizer can make use of them to do index-based join or sort-merge join. They are quicker than nest-loop join. In parallel RDBMSs, hash join has become one of the most popular algorithms.

So, Norris, if you want to speed your join, you can do this:

1. switch to cost-based optimization option, if there is.
2. create indexes on some columns
3. avoid natural-join if possible.

Sent via Deja.com http://www.deja.com/
Before you buy. Received on Tue Jun 13 2000 - 00:00:00 CDT