Skip site navigation (1)Skip section navigation (2)
Date:      Sun, 8 Sep 1996 04:04:56 -0400 (EDT)
From:      Brian Tao <taob@io.org>
To:        Chuck Robey <chuckr@glue.umd.edu>
Cc:        FREEBSD-HACKERS-L <freebsd-hackers@freebsd.org>, FREEBSD-SCSI-L <freebsd-scsi@freebsd.org>
Subject:   Re: Streamlogic RAIDION drive arrays
Message-ID:  <Pine.NEB.3.92.960908034759.14220N-100000@zap.io.org>
In-Reply-To: <Pine.OSF.3.95.960907185744.30216K-100000@fiber.eng.umd.edu>

next in thread | previous in thread | raw e-mail | index | archive | help
On Sat, 7 Sep 1996, Chuck Robey wrote:
>
> This seems somewhat odd to me.  I thought you'd expend the additional cost
> of RAID where you really need reliability.

    True enough... I was approaching a hardware RAID solution mainly
from a performance standpoint.  The need for a reliable e-mail server
is obvious, but people often mention how news is "expendable".  I
disagree with that.  A news server is just like a giant database
server:  it needs to be fast, big, *and* reliable.  I have five
non-customer feeds that consistently accept 90% of the articles I send
then, and another four above 50%.  If I lose even half a day's batch,
I'm going to hear about it.  Not to mention customers who get just as
upset about losing articles from their favourite groups as if it were
personal e-mail.

> If you lose news (assuming this doesn't happen very often) you just
> reload from another server, right?

    Assuming that some of your upstream feeds have spooled output
batches and can backfill you the missing articles.

> If you lose mail, it's gone, no backup at all.

    I'll be doing nightly backups of the mail spool to tape in case of
catastrophic failure.  With a projected 20GB of mail, this is going to
get rather expensive. :(  I read a study in some computer journal a
few months back that warned of a 2x to 5x increase in cost to make an
"normal" non-redundant system into a fully-redundant, high
reliability, self-monitoring and self-healing server.  I can start to
see why now, just in the disk subsystem alone.
--
Brian Tao (BT300, taob@io.org, taob@ican.net)
Senior Systems and Network Administrator, Internet Canada Corp.
"Though this be madness, yet there is method in't"




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?Pine.NEB.3.92.960908034759.14220N-100000>