RE: DB RAID Setup

  • From: "Crisler, Jon" <Jon.Crisler@xxxxxxx>
  • To: <martinfbrown@xxxxxxxxxxx>, <rob.dempsey@xxxxxxxxxx>, <oracle-l@xxxxxxxxxxxxx>
  • Date: Tue, 9 Dec 2008 19:12:57 -0500

Rob, I too agree that putting Data on Raid 5 and Index on Raid 10 is a
bit odd- I would either put them both on Raid 5, or both on Raid 10.
Since Indexes can be very large you are using a lot of storage to
maintain them.  Everything else I tend to agree with, but most of this
is a judgment call.   If you are trying to save on raw storage with data
on Raid 5, it would seem that you want to do the same with Indexes.  

 

Martin, we too use LOTS of 3par storage, and most of the time there is
little or no performance difference between Raid levels, until the cache
is saturated with writes.  You are already protecting your data with
some form of RAID, so doing another mirror (i.e. redo, archive etc.)
with Oracle is done to protect you against logical failures, which I
have seen with corrupted archive logs.  3Par certainly can be considered
state of the art, but then so is EMC and all the others.  We have tested
3par against other vendors such as HP EVA and frequently the EVA will
outperform 3par, but with 3par you get Thin Provisioning etc.  There is
nothing really "magic" about 3par- they are subject to the same
performance considerations as any other disk vendor with the added
advantage of Thin Provisioning. 

 

Also, find out if you have Snapshots occurring on a regular basis- this
can cause short spikes of high disk usage - by short I mean 5 to 60
seconds.  Also find out if you SAN administrator periodically rebalances
the frames (its like a big giant defrag).

 

You could very well be encountering poor performance due to some other
system on the 3Par frame competing with your activity.  Your SAN admin
should be able to tell you some performance metrics, 3par cache
utilization etc.  Ask him if he is using the 3Par System Reporter to
track performance data.   You can use AWR to look at your times of poor
performance and check the I/O service times against those times where
performance is better.  If you are typically returning 5ms service times
but with spikes at 10 or 15ms, this is a good thing to show the SAN guy.

 

________________________________

From: oracle-l-bounce@xxxxxxxxxxxxx
[mailto:oracle-l-bounce@xxxxxxxxxxxxx] On Behalf Of Martin Brown
Sent: Tuesday, December 09, 2008 4:17 PM
To: rob.dempsey@xxxxxxxxxx; oracle-l@xxxxxxxxxxxxx
Subject: RE: DB RAID Setup

 

I have a similar question regarding raid 5 and what I perceive to be a
"potential" problem. I am experiencing a high number of logfile sync
waits that appear most afternoons as the daily load starts to increase.
We're running 10.2.0.3 RAC with a 8 node cluster. We use a 3par storage
system for the database. My architect has configured both raid 1 and
raid 5 diskgroups and he duplexes the online redo logs and controlfiles
to raid5. 
 
Now, he insists that the logfile sync waits are not related in any way
to the raid5 configuration and the 3par storage solution is
state-of-the-art. But the only remedy I can find is to move to faster
SSD. Is it possible that the high afternoon activity coupled with the
raid5 configuration is the root of my problem? 
 
I've looked at other stuff too. The log switches are 2 - 3 an hour. The
log buffer is never full. Anyone have any other suggestions as to where
to look?



________________________________


Subject: DB RAID Setup
Date: Tue, 18 Nov 2008 17:06:20 +0000
From: Rob.Dempsey@xxxxxxxxxx
To: oracle-l@xxxxxxxxxxxxx



Hi

 

Oracle 10g2 (data warehouse)

 

I thought I would ask peoples' thoughts on the following.

 

I have setup our database whereby the index tablespace and data
tablespace are separate.  This is not for performance reason only for
ease of maintenance.  

 

We are being advised by the SAN provider to use the following RAID
layout

 

                Archive redo Logs            - RAID 10

Redo Logs                           - RAID 10

Temp tablespace             - RAID 10

Undo tablespace              - RAID 10

Index tablespaces           - RAID 10

System tablespace          - RAID 5

Data tablespaces              - RAID 5

 

Redo logs / Temp tablespace I agree with. 

 

To use RAID 5 for data, I understand there is a write performance hit
but this is a data warehouse so should be ok (Ideally I would like that
RAID 10 as well). But to have the index tablespace on RAID 10 and data
tablespace on RAID 5 I found that strange.  When I asked the reason why
I was give the response 'that is what Oracle recommends'. 

 

Has anyone heard this before?

 

Rob

 

 

 

 

 

 

________________________________

You live life online. So we put Windows on the web. Learn more about
Windows Live <http://clk.atdmt.com/MRT/go/127032869/direct/01/> 

Other related posts: