                                                                                                                                                                                                                                                                
Path: pr3ni23533pbb.0!nntp.google.com!news2.google.com!postnews.google.com!glegroupsg2000goo.googlegroups.com!not-for-mail
From: andrew.prota...@gmail.com
Newsgroups: comp.databases.oracle.server
Subject: Re: DBWR performance
Date: Sun, 13 May 2012 21:04:38 -0700 (PDT)
Organization: http://groups.google.com
Lines: 48
Message-ID: <14549587.3006.1336968278729.JavaMail.geo-discussion-forums@vbbhi10>
References: <a57eb2b1-e9a7-4a96-b92c-0d93418ca753@p6g2000yqi.googlegroups.com>
 <jo299h$paj$1@dont-email.me> <a5f98f06-9cf7-4292-b432-96844e7945c9@x17g2000yqj.googlegroups.com>
 <jo5bjd$sa8$1@dont-email.me> <29ced5ee-a0c4-4032-a082-32401b1dd1bb@35g2000yqq.googlegroups.com>
 <jo8f27$o78$1@dont-email.me> <c80b0a21-af6d-47e0-986a-658372c0bc1d@kw17g2000pbb.googlegroups.com>
 <b95d37bf-30f2-4dd7-a2a4-6eb20189b95e@pw2g2000pbb.googlegroups.com>
 <20767615.573.1336577515844.JavaMail.geo-discussion-forums@ynff7>
 <31713404.61.1336787499050.JavaMail.geo-discussion-forums@ynmk20> <3e7c204c-2134-48ff-be9d-5b4c1d80ce21@h10g2000pbi.googlegroups.com>
NNTP-Posting-Host: 99.158.173.169
Mime-Version: 1.0
NNTP-Posting-Date: Mon, 14 May 2012 04:04:39 +0000 (UTC)
In-Reply-To: <3e7c204c-2134-48ff-be9d-5b4c1d80ce21@h10g2000pbi.googlegroups.com>
Complaints-To: groups-abuse@google.com
Injection-Info: glegroupsg2000goo.googlegroups.com; posting-host=99.158.173.169;
 posting-account=k-XxUwoAAADQPimkMV_wzL7XsuppPEFh
User-Agent: G2/1.0
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: quoted-printable

Adaptec is in and here are some results with raid1:

1. Write speeds:

lgwr - 130MB/sec
dbwr - 150MB/sec
direct path - 160MB/sec

Still no cigar :-) (200).

2. Lgwr looks like limited by background process running insert select and =
using 100% cpu, but if I do 2 sqlplus processes doing the same concurrently=
 then lgwr speed jumps to 150MB/sec briefly and drops back to 130MB/sec. No=
ne of 3 processes is doing 100% CPU, so there is some other bottleneck. May=
be just shared memory access and waiting for semaphores.

3. Bonnie++ shows write/read 180/330 MB/sec vs 180/180 MB/sec with mdadm. T=
his means adaptec knows little trick called reading from 2 hdd in raid 1 co=
ncurrently and mdadm does not know it.

4. If you (adaptec) are designing pc adapter and know that it is overheatin=
g then you DO NOT write everywhere in the doc that you operating temperatur=
e tops at 55C. You take big radiator or small fan or both and mount them on=
 your PCI card, as nvidia or ati does on all their graphic cards for last g=
azillion years. Customer must not be forced to put extra fan to cool your c=
ard.

5. The same goes for big capacitor (replaces battery) hanging of the card o=
n twisted wire - it is not my problem to figure out where to mount it in AT=
 box - this has to be integrated in PCI card.

Andrew

> > 6. this is a winning combo so far (1 writer, no slaves, direct io):
> >
> > db_writer_processes=3D1
> > filesystemio_options=3Ddirectio
> > log_buffer=3D81920000
> >
> > Constant write speed
> >
> > lgwr - 130MB/sec
> > dbwr - 100MB/sec
>=20
> Interesting the large log buffer.  Any thoughts/results on/with other
> sizes?

It does not matter for lgwr 8m or 80M - it is about the same.