Re: DBWn performs batched (multiblock) writes to improve efficiency

From: Tanel Põder <tanel.poder.003_at_mail.ee>
Date: Wed, 21 Apr 2004 23:57:40 +0300
Message-ID: <045b01c427e3$49113360$14f823d5@porgand>

> Hi, Tanel:
> I did a test again. I tried both batch delete, and checkpoint. Test
> result shows it is always single block write. Maybe My test is limited,
but
> I did not see multiblockk write.

Hm... maybe it's a platform issue..? Or something to do with asynch IO (that Oracle leaves the write coalescing to OS instead of doing it by itself?). Just guessing here...

I tested with 10g on linux and saw up to 112k coalesced writes (that was the maximum in my system), however I couldn't test this on 9.2 nor 10g beta on my windows (the _dbwr_tracing seems to behave differently in these and I don't have any utility to trace IO system calls in windows).

Btw, what are the values for your _db_writer_coalesce_write_limit and _db_writer_coalesce_area_size parameters?

Tanel.

> Can you verify:
>
> --solaris 8,vxvm raw datafile, and ufs datafile.
> [oracle_at_main-db6**testdb]$cat dbwr_write.log |grep "kaio(AIOWRITE" |awk
> '{print $4}' |uniq
> 8192,
>
> I also used vxstat -g rootdg -i 2 -c 1000, write the output to a logfile.
> and ,
> 160
> [root_at_V480 /]# cat vx.log | awk '{print $4, $2}' |grep -v "0 0" |awk
'{print
> $1/$2}'|uniq
> 16

Please see the official ORACLE-L FAQ: http://www.orafaq.com

To unsubscribe send email to: oracle-l-request_at_freelists.org put 'unsubscribe' in the subject line.

--
Archives are at http://www.freelists.org/archives/oracle-l/
FAQ is at http://www.freelists.org/help/fom-serve/cache/1.html
-----------------------------------------------------------------

Received on Wed Apr 21 2004 - 15:54:49 CDT