RE: Oracle on RAID - IOwait problem

Calvin,

This is a common RAID misunderstanding which starts off as RAID IS GOOD.

Oracle recommend using multiple disks to spread the load between the different file access profiles for data, index, system, rollback, log and temp. To avoid queueing/contention problems.

What you had is Oracle on ONE disk, which may well be up to 25% slower for writes and 25% faster on reads but it is still all on the ONE disk. Basically, your reads will be slowed by the writes.

Summary?
One RAID 5 'spindle' is a killer. You need at least 3 intependant RAID 5 spindles thats 3 x 5 drives per spindle and not logical partitioning.

George.

> -----Original Message-----
> From: "Calvin Lim" <calvin-lim_at_mediaring.com>
> [mailto:calvin-lim_at_mediaring.com]
> Sent: Friday, May 05, 2000 7:55 AM
> To: smtp_at_inl001@servers["Multiple recipients of list ORACLE-L"
> <ORACLE-L_at_fatcity.com>]
> Subject: Oracle on RAID - IOwait problem
>
>
> Hi,
>
> We've just put a 8.1.5 on a 2CPU Sun Solaris into production.
>
> The machine is hooked on to a RAID5 storage device by Seek.
> Oracle datafiles are spread across 3 channels on the RAID.
> What i'm noticing is that the disk iowait is reaching incredible
> levels and top is reporting 0.0% idle for CPU states and 100M
> swaps.
>
> We're currently suspecting that it could be 1. the RAID
> configuration thats giving problem, 2. some cron driven perl/DBI
> jobs that is pounding the Database too often too hard with
> inefficient queries.
>
> Could 2. be a reason for extensive iowaits thats always hovers
> round the 85% mark. I need some advise on how one would go about
> troubleshooting such iowait related problems? Thanks very much.
>
> Regards,
> Calvin
>
> --------------------
>
> load averages: 0.39, 0.43, 0.46 22:19:30
> 74 processes: 71 sleeping, 1 running, 2 on cpu
> CPU states: 0.0% idle, 5.2% user, 8.5% kernel, 86.3%
> iowait, 0.0% swap
> Memory: 1024M real, 16M free, 100M swap in use, 1900M swap free
>
> PID USERNAME THR PRI NICE SIZE RES STATE TIME CPU COMMAND
> 2359 root 1 59 0 1944K 1080K sleep 0:03 2.40% sshd1
> 29216 oracle 1 58 0 832M 799M cpu0 1:50 2.12% oracle
> 2383 oracle 1 58 0 2000K 1360K cpu2 0:00 1.24% top
> 29214 mrurl 1 53 2 13M 4840K run 0:54 0.98%
> ifmxTrigger
> 10591 oracle 1 60 0 834M 798M sleep 102:23 0.95% oracle
> 27178 oracle 1 38 0 2000K 1160K sleep 0:55 0.51% top
> 10360 oracle 1 60 0 832M 798M sleep 113:10 0.50% oracle
> 10362 oracle 1 60 0 832M 797M sleep 112:00 0.49% oracle
> 10366 oracle 1 60 0 832M 797M sleep 111:50 0.42% oracle
> 10364 oracle 1 60 0 832M 798M sleep 112:27 0.41% oracle
> 10368 oracle 1 60 0 832M 797M sleep 112:16 0.38% oracle
> 2389 oracle 1 60 0 831M 799M sleep 0:00 0.19% oracle
> 10338 oracle 1 59 0 833M 794M sleep 3:01 0.15% oracle
> 10340 oracle 1 58 0 832M 794M sleep 4:40 0.14% oracle
> 9341 oracle 1 48 0 8960K 2352K sleep 9:34 0.13% tnslsnr
>
> ----------------
> --
> Author: Calvin Lim
> INET: calvin-lim_at_mediaring.com
>
> Fat City Network Services -- (858) 538-5051 FAX: (858) 538-5051
> San Diego, California -- Public Internet access / Mailing Lists
> --------------------------------------------------------------------
> To REMOVE yourself from this mailing list, send an E-Mail message
> to: ListGuru_at_fatcity.com (note EXACT spelling of 'ListGuru') and in
> the message BODY, include a line containing: UNSUB ORACLE-L
> (or the name of mailing list you want to be removed from). You may
> also send the HELP command for other information (like subscribing).
Received on Fri May 05 2000 - 03:48:51 CDT