Re: ADDM advices Oracle's application logic

On 20 jan, 08:49, Helma <helma.vi..._at_hotmail.com> wrote:
> On 19 jan, 21:50, Helma <helma.vi..._at_hotmail.com> wrote:
>
>
>
> > On 19 jan, 15:25, Frank van Bortel <frank.van.bor..._at_gmail.com> wrote:
>
> > > Helma wrote:
> > > > On 18 jan, 19:16, Frank van Bortel <frank.van.bor..._at_gmail.com> wrote:
> > > >> Frank van Bortel wrote:
> > > >>> File size set to 1048576 KB
> > > >>> SYNC Mode.
> > > >> [snip!]
>
> > > >> Now, lets try async:
> > > >> [root_at_cs-frank03 ~]# iozone -a -s 1G
>
> > > >> KB reclen write rewrite read reread
> > > >> 1048576 4 504582 1146300 1713216 1763957
>
> > > >> random random bkwd record stride
> > > >> read write read rewrite read
> > > >> 1366258 1029497 1445332 1212665 1336920
>
> > > >> fwrite frewrite fread freread
> > > >> 483586 1026453 1101598 1657060
>
> > > >> --
>
> > > >> Regards,
> > > >> Frank van Bortel
>
> > > >> Top-posting in UseNet newsgroups is one way to shut me up
>
> > > > hi Frank,
>
> > > > OS and IO is a bit new to me, i didn't know the iozone tool, i hope to
> > > > give the correct interpretation.
> > > > On my system:
>
> > > > ./iozone -a -s 1G -b xcel
>
> > > > gave the following on the 4kb recordsize:
>
> > > > Writer Report 244166
> > > > Re-writer Report 681441
> > > > Reader Report 1340226
> > > > Re-reader Report 1384200
> > > > Random Read Report 1066901
> > > > Random Write Report 586944
> > > > Backward Read Report 1222318
> > > > Record Rewrite Report 1263385
> > > > Stride Read Report 1135547
> > > > Fwrite Report 234427
> > > > Re-fwrite Report 616447
> > > > Fread Report 1297864
> > > > Re-fread Report 1332997
>
> > > > The write seems half of yours ( 504582 vs 244166). Rather big
> > > > difference, and it seems you've got the slower hardware too...
> > > > H.
>
> > > As long as we both use the same options, results should
> > > be comparable, isn't it?
>
> > > Slow hardware, me?!? I'm pretty content with it, I can tell you.
> > > How much memory do you have installed, and what kernel do you run?
> > > [oracle102_at_cs-frank03 mrca]$ uname -a
> > > Linux cs-frank03 2.6.18-53.1.4.el5PAE #1 SMP Fri Nov 30 01:21:20 EST
> > > 2007 i686 i686 i386 GNU/Linux
> > > 4GB installed (hence the PAE kernel - 32bit, though)
>
> > > Someone should start a database on this.
> > > --
>
> > > Regards,
> > > Frank van Bortel
>
> > > Top-posting in UseNet newsgroups is one way to shut me up
>
> > Hi Frank,
>
> > My machine is a 4 CPU 8Gb HP Proliant ML 570 64bits Linux. I'll run
> > the uname if i can get in the office tomorrow.
> > HV
> > ps:
> > I said your hardware was slower, not slow.
>
> uname -a reveals:
>
> [root_at_rafiki ~]# uname -a
> Linux rafiki 2.6.9-5.ELsmp #1 SMP Wed Jan 5 19:29:47 EST 2005 x86_64
> x86_64 x86_
> 64 GNU/Linux
>
> > > >>> SYNC Mode.
> > > >> [snip!]
>
> > > >> Now, lets try async:
>
> So there IS a go_faster = TRUE parameter ;-)

BTW i use raid 0+1 Received on Sun Jan 20 2008 - 01:51:02 CST