90 lines
4.1 KiB
Plaintext
90 lines
4.1 KiB
Plaintext
From fork-admin@xent.com Tue Sep 17 11:29:50 2002
|
|
Return-Path: <fork-admin@xent.com>
|
|
Delivered-To: yyyy@localhost.example.com
|
|
Received: from localhost (jalapeno [127.0.0.1])
|
|
by jmason.org (Postfix) with ESMTP id DB4C516F03
|
|
for <jm@localhost>; Tue, 17 Sep 2002 11:29:49 +0100 (IST)
|
|
Received: from jalapeno [127.0.0.1]
|
|
by localhost with IMAP (fetchmail-5.9.0)
|
|
for jm@localhost (single-drop); Tue, 17 Sep 2002 11:29:49 +0100 (IST)
|
|
Received: from xent.com ([64.161.22.236]) by dogma.slashnull.org
|
|
(8.11.6/8.11.6) with ESMTP id g8GJhcC07013 for <jm@jmason.org>;
|
|
Mon, 16 Sep 2002 20:43:39 +0100
|
|
Received: from lair.xent.com (localhost [127.0.0.1]) by xent.com (Postfix)
|
|
with ESMTP id 89BE82940D0; Mon, 16 Sep 2002 12:40:05 -0700 (PDT)
|
|
Delivered-To: fork@example.com
|
|
Received: from venus.phpwebhosting.com (venus.phpwebhosting.com
|
|
[64.29.16.27]) by xent.com (Postfix) with SMTP id B6F8B2940CD for
|
|
<FoRK@xent.com>; Mon, 16 Sep 2002 12:39:05 -0700 (PDT)
|
|
Received: (qmail 8446 invoked by uid 508); 16 Sep 2002 19:43:47 -0000
|
|
Received: from unknown (HELO hydrogen.leitl.org) (62.155.144.96) by
|
|
venus.phpwebhosting.com with SMTP; 16 Sep 2002 19:43:47 -0000
|
|
Received: from localhost (eugen@localhost) by hydrogen.leitl.org
|
|
(8.11.6/8.11.6) with ESMTP id g8GJBlB15944; Mon, 16 Sep 2002 21:11:47
|
|
+0200
|
|
X-Authentication-Warning: hydrogen.leitl.org: eugen owned process doing -bs
|
|
From: Eugen Leitl <eugen@leitl.org>
|
|
To: <sdw@lig.net>
|
|
Cc: Gary Lawrence Murphy <garym@canada.com>,
|
|
Udhay Shankar N <udhay@pobox.com>, Adam Rifkin <adam@KnowNow.com>,
|
|
<FoRK@xent.com>
|
|
Subject: Re: storage bits
|
|
In-Reply-To: <3D862A52.1080502@hpti.com>
|
|
Message-Id: <Pine.LNX.4.33.0209162104390.14309-100000@hydrogen.leitl.org>
|
|
MIME-Version: 1.0
|
|
Content-Type: TEXT/PLAIN; charset=US-ASCII
|
|
Sender: fork-admin@xent.com
|
|
Errors-To: fork-admin@xent.com
|
|
X-Beenthere: fork@example.com
|
|
X-Mailman-Version: 2.0.11
|
|
Precedence: bulk
|
|
List-Help: <mailto:fork-request@xent.com?subject=help>
|
|
List-Post: <mailto:fork@example.com>
|
|
List-Subscribe: <http://xent.com/mailman/listinfo/fork>, <mailto:fork-request@xent.com?subject=subscribe>
|
|
List-Id: Friends of Rohit Khare <fork.xent.com>
|
|
List-Unsubscribe: <http://xent.com/mailman/listinfo/fork>,
|
|
<mailto:fork-request@xent.com?subject=unsubscribe>
|
|
List-Archive: <http://xent.com/pipermail/fork/>
|
|
Date: Mon, 16 Sep 2002 21:11:47 +0200 (CEST)
|
|
X-Spam-Status: No, hits=-11.1 required=7.0
|
|
tests=EMAIL_ATTRIBUTION,IN_REP_TO,KNOWN_MAILING_LIST,
|
|
QUOTED_EMAIL_TEXT,SPAM_PHRASE_00_01,USER_AGENT_PINE,
|
|
X_AUTH_WARNING
|
|
version=2.50-cvs
|
|
X-Spam-Level:
|
|
|
|
On Mon, 16 Sep 2002, Stephen D. Williams wrote:
|
|
|
|
> It's efficient-end, not low end. At 1Million hour MTBF, 133MB/sec,
|
|
> and pretty good buffering and speed, the only thing going for SCSI is
|
|
> 15,000 RPM vs. 7200 and in a very small number of cases, slightly
|
|
> better scatter-gather. (Actually, I think there might be a 15,000 RPM
|
|
> IDE now.)
|
|
|
|
It's not just krpm, the desktop HDs have a higher failure rate. But I
|
|
agree, EIDE has high density, and EIDE hardware RAID can offer SCSI a
|
|
sound beating for reliability, performance, and storage density/rack units
|
|
for the money, if designed for it, and if people would actually start
|
|
buying it.
|
|
|
|
> The other issues are pretty much non-issues: using multiple drives and
|
|
> controller contention (just use many IDE channels with extra PCI
|
|
> cards, up to 10 in some systems), and long cable runs (just split
|
|
|
|
There are not all that many hard drives inside an 1U enclosure. Airflow
|
|
blockage (you have to fit in 2-3x the number of SCSI disks with EIDE) will
|
|
soon be a thing of the past due to SATA.
|
|
|
|
> storage between nodes). Dual-port SCSI is also a non-issue since it
|
|
> is very expensive, doesn't work that well in practice because there
|
|
> are numerous secondary failure modes for shared disk systems, and
|
|
> because you still end up with a single point of failure.
|
|
|
|
Since rack-space costs dominate, and our systems need more or less decent
|
|
I/O we're going with 1U Dells with SCSI. The hard drive prices don't
|
|
really make a visible difference, given the cost of the iron, and the
|
|
rackspace/month. Plus, 1U Dells don't have any space left for lots of EIDE
|
|
drives.
|
|
|
|
|