Oracle FAQ Your Portal to the Oracle Knowledge Grid
HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US
 

Home -> Community -> Mailing Lists -> Oracle-L -> RE: Hitachi 9000 (was RE: 2GB or not 2GB )

RE: Hitachi 9000 (was RE: 2GB or not 2GB )

From: Allen, Brandon <Brandon.Allen_at_OneNeck.com>
Date: Mon, 11 Sep 2006 15:56:14 -0700
Message-ID: <04DDF147ED3A0D42B48A48A18D574C45059E1EE1@NT15.oneneck.corp>


Here is a problem I had in a similar situation - our Netbackup server was using some space on a SAN for its disk staging pool (before writing backups to tape). Some of our databases were on the same SAN and would have terrible I/O performance from time to time - turned out it was because we were relying heavily on the SAN cache since we were running RAID5, and when the backups to disk would run, they would consume all the cache on the SAN and force the databasee to wait on the terribly slow RAID5 writes to disk. The solution was to disable caching for the backups on the SAN.

Lessons learned: beware of RAID5 (as you seem to be already); beware of your dependency on cache if you decide to use RAID5 despite the dangers; and, beware of sharing disk subsystems - you never know who/what might cause your database performance to take a dive.  

> -----Original Message-----
> From: oracle-l-bounce_at_freelists.org
> [mailto:oracle-l-bounce_at_freelists.org] On Behalf Of Keith Moore
> Sent: Monday, September 11, 2006 3:10 PM
> To: kevinc_at_polyserve.com
> Cc: oracle-l
> Subject: RE: Hitachi 9000 (was RE: 2GB or not 2GB )
>
> I check my data and found it is actually 102 GB in 66 minutes
> for a rate of 26 MB/sec for a full table scan. "Sequential
> reads" are much slower, around 8-10 MB/sec.
>
> Here is what I know about the architecture:
> Two Hitachi SANs
> 2 SANs are shared by 4 Sun 15K servers
> Disks are 146 GB each
> Each disk split into 35 GB Logical Volumes
> 8 disks in a logical disk group (Raid 5 w/ 128K stripe)
> 2 HBA per SAN. I think 1Gb, but not sure
> Veritas file system
> 56 GB read/write cache
>
> I know Raid10 would be better, but Raid5 is the corporate
> direction. The other problems seem to be that the disks are
> shared by multiple databases and possibly the stripe size.
>
>
> Any comments would be appreciated.
>
> Keith
>
> >
> > That is 113MB/s. You wouldn't happen to have 1Gb HBAs would you?
> > Is this data striped across, say, 10 spindles?
> >
>
>
> --
> http://www.freelists.org/webpage/oracle-l
>
>
>

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.

--
http://www.freelists.org/webpage/oracle-l
Received on Mon Sep 11 2006 - 17:56:14 CDT

Original text of this message

HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US