Re: now what??
- From: Robyn <robyn.sands@xxxxxxxxx>
- To: oracle-l@xxxxxxxxxxxxx
- Date: Tue, 24 Aug 2004 12:35:01 -0400
Hi Darrell,
The sysadmins are using Glance, but mentioned maybe getting the EMC
tools just this morning. The files that seem to be most affected by
the problem waits are actually our smaller tables (4 to 6 gb), so
breaking them up seemed unlikely to help. None of the datafiles are
over 2 gb anyway, and many are in smaller pieces (which bugs the heck
out of my OCD side - really messy). I do have Jonathon's book along
side Cary's, and have included system stats in my wish list of
changes, but I was waiting until the 9.2.0.5 patch went into
production to try them out. Our test warehouse has been patched and
it's performing fairly well - but it only has 2 cpus and one quarter
of the memory.
Nothing like living the experience to bring the message home.
Robyn
On Tue, 24 Aug 2004 08:46:16 -0700 (PDT), darrell@xxxxxxxxxxx
<darrell@xxxxxxxxxxx> wrote:
> Robyn,
>
> I'd be interested in knowing where the sysadmins are
> seeing drive queueing. If they're seeing it in Glance,
> I don't think this is as much file size / layout as
> potentially an overworked path(s) between the HP box
> and the EMC chassis. If it's on EMC monitoring tools,
> that's a different factor, but to my knowledge, Glance
> can't see beyond the OS and local hardware.
> How big are the files they want you to break up?
> If there is true disk queuing, then I'd ask what they
> are doing to move 'hot' file systems to hypers on
> lesser utilized disks. (EMC has tools to do this on
> the fly, I'd inquire to see if that is available in
> your environment.)
> Also, given the increase in CPU count and speed, I
> wonder if you might not benefit from system statistics
> (if you're not already gathering them). Take a look at
> Jonathon Lewis' (very informative) paper at:
> http://www.oracle.com/technology/pub/articles/lewis_cbo.html
>
> I'd also push forward with that 9.2.0.5 patch. You've
> already seen that it can make a difference as well, I'd
> be concerned that any work (tuning) you do now would
> not behave the same after patching.
>
> Yes, yes, yes to locally managed tablespaces. I don't
> have any handy evidence of performance improvement, but
> they are definitely easier administratively.
>
> One more thought on the large files, if there is a lot
> of concurrent use, you may benefit from multiple
> smaller files. If you're on filesystems and not raw
> devices, HP (and probably others) has some factor with
> single use locking from the OS. I think this also is
> in a Jonathon Lewis paper out there somewhere.
>
> Good luck!
>
>
> ----------------------------------------------------------------
> Please see the official ORACLE-L FAQ: http://www.orafaq.com
> ----------------------------------------------------------------
> To unsubscribe send email to: oracle-l-request@xxxxxxxxxxxxx
> put 'unsubscribe' in the subject line.
> --
> Archives are at //www.freelists.org/archives/oracle-l/
> FAQ is at //www.freelists.org/help/fom-serve/cache/1.html
> -----------------------------------------------------------------
>
----------------------------------------------------------------
Please see the official ORACLE-L FAQ: http://www.orafaq.com
----------------------------------------------------------------
To unsubscribe send email to: oracle-l-request@xxxxxxxxxxxxx
put 'unsubscribe' in the subject line.
--
Archives are at //www.freelists.org/archives/oracle-l/
FAQ is at //www.freelists.org/help/fom-serve/cache/1.html
-----------------------------------------------------------------
Other related posts:
