Re: PQO and the rule-based optimizer

There are two options I remember of:

1- You can define your table with a degree of parallelism in your CREATE TABLE
2- You can specify it with HINTS in your query.

Our product, TunaSQL for Oracle is a SQL optimizer and supports parallel queries. You can get a demo at www.cardume.com

Best Regards

Fernando (morgan_at_cardume.com)
CardumeSoftware
www.cardume.com

jim nash <jimnash_at_sprintmail.com> wrote in article <3357C001.5D58_at_sprintmail.com>...
> One of the databases I'm working with has a large (3GB) table
> that must be scanned (full table scans) for a number of queries.
> Assuming sufficient cpu and disk resources, I understand that
> that the Parallel Query Option (PQO) can be used to divide
> the work among a number of servers, reducing query elapsed times.
>
> Oracle documentation leads me to believe that the query plan
> chosen for such operations relates to the distribution of data
> in the table, implying cost-based optimization.
>
> The vendor of this software insists on the use of the rule-based
> optimizer. We have specified such in init.ora, but we
> and also analyze all tables periodically for maintenance info and
> for "hint" optimizations (where the cost-based optimizer is used).
>
> The question is how does Oracle determine how to partition
> the query? Is the cost-based optimizer used under the sheets?
> Or is the table simply divided according to the number of
> data blocks?
>
> Oracle 7.1.6 is being used with HP/UX 10.10. The machine has
> 10 processors.
>
> Thanks,
>
> - Jim
>
>
Received on Mon Apr 21 1997 - 00:00:00 CDT