Re: Performance problems with large table

Hello,

I think that

        SELECT COUNT(ROWID)
will perform better than

        SELECT COUNT(*) Regards,
Igor

Ing. Martin Gamperl wrote:
>
> I created one table with 500 columns and about 84600 records. The first
> column is of type DATE the remaining comlumns are of type NUMBER.
> Columns are added or deleted from the table using triggers when entries
> in other tables change.
> The database was created with a block size of 8K under Windows NT 4.0
> with Oracle Enterprise Server 8.0.3.
> When performing queries response time is very poor! For example a
> "SELECT COUNT(*) ..." needs about 5 minutes to complete.
> At last I used the ANALYZE command with the LIST CHAINED ROWS option to
> determine chained or migrated rows and almost every second row is a
> chained row. So I copied the contents of the whole table to a new table
> to avoid chained rows - but with no success.
> What can we do to increase performance and decrease chained rows. Should
> we change initialization parameters (db_block_size, etc.)?
>
> Please reply to my email address.
>
> Thanxs in advance
>
> Martin
>
> --
> ***************************************************
> * Ing. Martin Gamperl
> * MGS Messgeraetesteuerung - Softwareentwicklung
> * Klosterwiesgasse 31/IV
> * A-8010 GRAZ
> * Tel. : +43-316-83 58 26
> * Fax : +43-316-83 58 26
> * Mobil : +43-664-40 03 032
> * EMail : mgampi_at_ibm.net
> ***************************************************

--
Igor Sereda,
ITC, Russia