Skip site navigation (1)Skip section navigation (2)
Date:      Mon, 31 Jan 2000 10:48:28 +1030
From:      Greg Lehey <grog@lemis.com>
To:        Marc Tardif <admin@wtbwts.com>
Cc:        freebsd-scsi@FreeBSD.ORG
Subject:   Re: hardware vs software stripping
Message-ID:  <20000131104827.A62824@freebie.lemis.com>
In-Reply-To: <Pine.BSF.4.10.10001301401360.60037-100000@server.b0x.com>
References:  <Pine.BSF.4.10.10001301401360.60037-100000@server.b0x.com>

next in thread | previous in thread | raw e-mail | index | archive | help
On Sunday, 30 January 2000 at 14:24:54 +0000, Marc Tardif wrote:
> I've been reading the freebsd-scsi mailing list, the dpt.com website and
> searching all over the web to find a solution for performing fast reads on
> an array of disks. The best choice seems to be raid-0 for our database
> using postgresql. The alternatives are either raid-1 which seems too
> wasteful on disks or raid-5 which provides fault tolerance. This last
> option could substituted for a tape backup and the possibility of a few
> minutes down time in case of disk failure.

I'm not sure I understand this sentence.  Are you planning to forget
RAID-5 after all and use a tape backup?  For reasonably large disks,
your downtime will be measured in hours, not minutes.  For a RAID-5
array, you shouldn't get any down time.

> roblem is: hardware or software?
>
> For hardware, I've been looking at the smartraid iv controller from dpt
> and perhaps their smartstation7 for an array of 7 x (9 or 18 gig drives).
> Although this solution isn't out of our price range, it seems to be better
> suited for raid-5. Considering raid-5 does stripping at the byte level, it
> might not even be as fast as raid-0 which also saves on redundancy.

I suppose you mean striping.  RAID-5 doesn't stripe at the byte level,
it stripes at the block level.  RAID-3 stripes at the byte level.

Using RAID-0 "saves" on redundancy: if you lose a disk, you lose the
file system.  There are also some complex interactions between ufs and
striping which can make it more than useless.  I haven't done any real
life tests, but it's possible that having striping doesn't buy you
very much at the best of times when using ufs.

> Therfore, for something as simple as stripping, using such hardware
> could be overkill. Besides, I'll still have to use ccd to do the
> stripping of the disks for raid-0.

I don't understand this comment.  If you want to do striping, you have
the choice of hardware RAID, ccd or vinum.

> For software, I think freebsd's ccd could provide all the services
> required for very fast i/o on a straight array of disks and at a
> fraction of the cost of the raid alternative. Unfortunately, I have
> never witnessed the virtues of raid myself and I am not in a
> position to make an educated decision. I would therefore appreciate
> if someone from this mailing list could share their experience to
> help with my dilemna.

Why do you want to use ccd and not vinum?  In any case, you may find
that either are faster than the DPT controller (if you can find one;
they're no longer making them).

Greg
--
Finger grog@lemis.com for PGP public key
See complete headers for address and phone numbers


To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-scsi" in the body of the message




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20000131104827.A62824>