Re: Speeding up the redo logs

HI Everyone,

Thanks for writing. I've been struggling with this for a while now.

We are running VALinux. 8.1.7.0.0 Enterprise edition. Linux 2.2.14-VA.5.1smp #1 SMP Tue Sep 12 21:26:21 PDT 2000 i686 All configured with RAID 5. 4 CPUs

We initially began testing with sqlloader in parallel. And with partitioned tables. However, I soon noticed that the slowest part was the redo logs. So, this is what I am concentrating on now.

Any other ideas?

Thanks

In article <94akes$e02$1_at_nnrp1.deja.com>,   argosy22_at_my-deja.com wrote:
> Hi all,
>
> I'm currently working on a project where we will eventually
> want to insert up to a billion rows a day into the database.
> This is an average of over 11,000 rows per second.
>
> To this point, I've just made a table called JUNK:
>
> Name Null? Type
> ----------------------------------------- --------
> ----------------------------
> JUNK_ID NUMBER(6)
> JUNK_DESC VARCHAR2(20)
>
> I truncate this, and populate it with 1 million rows with a PLSQL
> routine. It takes 2 minutes 38 seconds to populate it.
> About 6,300 inserts per second. Naturally, if I add indexes, it
> slows the inserts down a lot.
>
> The main part that slows everything down is the redo logs.
> There are five 50 meg redo logs on their own drive.
>

Sent via Deja.com
http://www.deja.com/ Received on Mon Jan 22 2001 - 11:08:31 CST