Skip site navigation (1)Skip section navigation (2)
Date:      Wed, 17 Jan 2007 00:57:05 +0100
From:      Ivan Voras <ivoras@fer.hr>
To:        "Wood, Russell" <Russell.Wood@rac.com.au>
Cc:        freebsd-performance@freebsd.org, freebsd-questions@freebsd.org
Subject:   Re: Very Poor Raid Card Performance
Message-ID:  <45AD6651.5010107@fer.hr>
In-Reply-To: <D6B5375CBC761B4BAD78E414B4BB557266318C@mercury.rac.com.au>
References:  <D6B5375CBC761B4BAD78E414B4BB557266318C@mercury.rac.com.au>

next in thread | previous in thread | raw e-mail | index | archive | help
This is an OpenPGP/MIME signed message (RFC 2440 and 3156)
--------------enig21B123D9444D806EC45519D7
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

Wood, Russell wrote:

> I have an LSI MegaRaid with three drives at RAID 5: works a
> god-damn-treat. I don't know what my throughput is on RAID 5 but RAID 0=

> was fast (70MB/s+).

Mine was RAID5. Hmm, oops, it wasn't LSI but HP/Compaq CISS, battery
backed, SATA on SAS.

da0 at ciss0 bus 0 target 0 lun 0
da0: <COMPAQ RAID 5  VOLUME OK> Fixed Direct Access SCSI-0 device
da0: 135.168MB/s transfers
da0: 476886MB (976662896 512 byte sectors: 255H 32S/T 65535C)

> Were you doing these tests a few hours after initializing the array, by=

> any chance? If so, then the poor performance could be because the array=

> was `in build' while you were using it, and therefore you'd get very
> poor performance.

Yes, I observed BIOS status. Here's a typical run of bonnie++:

Version 1.93c       ------Sequential Output------ --Sequential Input-
--Random-
Concurrency   1     -Per Chr- --Block-- -Rewrite- -Per Chr- --Block--
--Seeks--
Machine        Size K/sec %CP K/sec %CP K/sec %CP K/sec %CP K/sec %CP
/sec %CP
xxx.com       2500M   358  99 13446   4  7068   2   559  96 97548  23
300.0  10
Latency             23236us     848ms     488ms     191ms     214ms
175ms

My reading of iostat during 'dd' is that writes perform good until the
on-controller cache gets filled, then it slows down:

      tty             da0            pass0             cpu
 tin tout  KB/t tps  MB/s   KB/t tps  MB/s  us ni sy in id
=2E..
   0   60   0.00   0  0.00   0.00   0  0.00   0  0  0  0 100
   1   64 127.93 217 27.08   0.00   0  0.00   0  0  5  0 95
   0   61 125.92 400 49.14   0.00   0  0.00   0  0  2  1 98
   0   61 128.00 107 13.36   0.00   0  0.00   0  0  2  0 98
   0   61 128.00 107 13.36   0.00   0  0.00   0  0  2  0 98
   0   61 128.00 104 12.99   0.00   0  0.00   0  0  2  0 98

Cache size is 128MB, 75% dedicated to write caching.


--------------enig21B123D9444D806EC45519D7
Content-Type: application/pgp-signature; name="signature.asc"
Content-Description: OpenPGP digital signature
Content-Disposition: attachment; filename="signature.asc"

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.4 (MingW32)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFFrWZYldnAQVacBcgRAnHEAJ4vXIxQL0kUSVnhyCCWH47HTisKjQCdFLBx
LLKpARmf0xI5udqqNOznODg=
=ajTN
-----END PGP SIGNATURE-----

--------------enig21B123D9444D806EC45519D7--



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?45AD6651.5010107>