Hi Tom,
You do not say that the log file disk is dedicated to the log files. If not,
it should be. Other than that, keeping your commit frequency low is the next
most important factor. Also, the optimal log buffer size for this type of
operation is about 160K on most systems.
Regards,
Steve Adams
In article <7ffs63$o7d$1_at_nnrp1.dejanews.com>,
Tom Lewis <telewis_at_my-dejanews.com> wrote:
> I have an Oracle 8.0.5 database where the datafiles resides on an 8 disk RAID
> and the log files on a separate single disk. The application loads several
> million records into a number of tables (with no constraints) and then a
> PL/SQL cursor is used to copy the data into a relational model, via a
> select/insert process, to identify constraint violations. During large data
> loads and the cursor select/insert process, disk I/O on the log files seems
> to be the performance-limiting factor.
>
> Would putting the log files on the RAID help performance, which is more
> important than resilience in this application? Also, does log file size
> affect performance (currently in “noarchivelog” mode to avoid further I/O
> problems but should start archiving in the future)?
>
> Any ideas gratefully received.
>
> --
> Tom Lewis
>
> -----------== Posted via Deja News, The Discussion Network ==----------
> http://www.dejanews.com/ Search, Read, Discuss, or Start Your Own
>
-----------== Posted via Deja News, The Discussion Network ==----------
http://www.dejanews.com/ Search, Read, Discuss, or Start Your Own
Received on Mon Apr 19 1999 - 15:11:52 CDT
