Re: perplexing performance on AIX

  • From: "John Smith" <john40855@xxxxxxxxx>
  • To: "David Barbour" <david.barbour1@xxxxxxxxx>
  • Date: Tue, 6 May 2008 17:04:33 -0500

JFS2 file system mounted with CIO
OS:  5300-05-06
FilesystemIO_options=setall
I dont know what HBA's are being used.

Top five timed events:

~~~~~~~~~~~~~~~~~~                                        wait   Call
Event                                 Waits    Time (s)   (ms)   Time Wait
Class
------------------------------ ------------ ----------- ------ ------
----------
PX Deq Credit: send blkd             59,803       6,103    102   79.1
Other
CPU time                                            515           7.0
direct path read                     23,833         292     11    4.0   User
I/O
db file sequential read             139,499         225      2    3.1   User
I/O
direct path read temp                 7,632          98     13    1.4   User
I/O
          -------------------------------------------------------------




On Tue, May 6, 2008 at 4:50 PM, David Barbour <david.barbour1@xxxxxxxxx>
wrote:

> John,
>
> What version of AIX, including maintenance level.  What is your
> filesystemio_options parameter?  What HBAs are you using?  Raw, or journaled
> filesystems?  JFS2?  How are they mounted?
>
> What do you see with iostat?
>
> On Tue, May 6, 2008 at 4:32 PM, John Smith <john40855@xxxxxxxxx> wrote:
>
> > Not that I want to discourage any potential comedians, but I do
> > understand the concept of idle waits.  However, the wait event seems
> > excessive, (its larger than any other timed event on the system, to include
> > CPU time), which makes me suspect that there is a problem.
> >
> >
> > On Tue, May 6, 2008 at 3:11 PM, Baumgartel, Paul <
> > paul.baumgartel@xxxxxxxxxxxxxxxxx> wrote:
> >
> > >  On Tue, May 6, 2008 at 2:21 PM, Allen, Brandon <
> > > Brandon.Allen@xxxxxxxxxxx> wrote:
> > >
> > > >  If you disabled parallellism, you couldn't get PX wait events.  You
> > > > might want to try with DOP of 1.  I'd also verify CIO is being used if 
> > > > you
> > > > aren't sure.
> > > >
> > >
> > >
> > > Reminds me of the old Henny Youngman joke:  "I went to the doctor.  I
> > > said 'Doc, it hurts when I do this.'  He said 'Don't do that.'"
> > >
> > > *Paul Baumgartel*
> > > *CREDIT SUISSE*
> > > Information Technology
> > > Prime Services Databases Americas
> > > One Madison Avenue
> > > New York, NY 10010
> > > USA
> > > Phone 212.538.1143
> > > paul.baumgartel@xxxxxxxxxxxxxxxxx
> > > www.credit-suisse.com
> > >
> > >
> > >  ------------------------------
> > > *From:* oracle-l-bounce@xxxxxxxxxxxxx [mailto:
> > > oracle-l-bounce@xxxxxxxxxxxxx] *On Behalf Of *John Smith
> > > *Sent:* Tuesday, May 06, 2008 3:29 PM
> > > *To:* Allen, Brandon; oracle-l@xxxxxxxxxxxxx
> > > *Subject:* Re: perplexing performance on AIX
> > >
> > > We really want to be able to use parallel.
> > >
> > >
> > > >  ------------------------------
> > > > *From:* John Smith [mailto:john40855@xxxxxxxxx]
> > > > *Sent:* Tuesday, May 06, 2008 11:57 AM
> > > > *To:* Allen, Brandon
> > > > *Subject:* Re: perplexing performance on AIX
> > > >
> > > >   Yes, we tried decreasing the DOP.   Didnt really affect the wait
> > > > event.  I am pretty sure the drives are already mounted w/ CIO.  At 
> > > > least
> > > > they are supposed to be.
> > > >
> > > > On Tue, May 6, 2008 at 1:20 PM, Allen, Brandon <
> > > > Brandon.Allen@xxxxxxxxxxx> wrote:
> > > >
> > > > >  Have you tried decreasing the degree of parallelism (or maybe
> > > > > even disabling it) to see if that helps?  Sometimes too 
> > > > > muchparallelism is
> > > > > a bad thing.
> > > > >
> > > > > You might also want to consider using concurrent IO if you're not
> > > > > already.
> > > > >
> > > > > Regards,
> > > > > Brandon
> > > > >
> > > > >
> > > > >
> > > > > Privileged/Confidential Information may be contained in this
> > > > > message or attachments hereto. Please advise immediately if you or 
> > > > > your
> > > > > employer do not consent to Internet email for messages of this kind.
> > > > > Opinions, conclusions and other information in this message that do 
> > > > > not
> > > > > relate to the official business of this company shall be understood as
> > > > > neither given nor endorsed by it.
> > > > >
> > > >
> > > >   Privileged/Confidential Information may be contained in this
> > > > message or attachments hereto. Please advise immediately if you or your
> > > > employer do not consent to Internet email for messages of this kind.
> > > > Opinions, conclusions and other information in this message that do not
> > > > relate to the official business of this company shall be understood as
> > > > neither given nor endorsed by it.
> > > >
> > >
> > > ==============================================================================
> > > Please access the attached hyperlink for an important electronic 
> > > communications disclaimer:
> > > http://www.credit-suisse.com/legal/en/disclaimer_email_ib.html
> > > ==============================================================================
> > >
> > >
> >
>

Other related posts: