Date: Mon, 30 Oct 2000 15:00:26 +0000 From: Steve Roome <steve@sse0691.bri.hp.com> To: David Kelly <dkelly@hiwaay.net> Cc: Soren Schmidt <sos@freebsd.dk>, Thierry.Herbelot@alcatel.fr, stable@FreeBSD.ORG Subject: Re: IDE drives to avoid ? [was : Re: ATA problems (rehashed)] Message-ID: <20001030150026.B14579@moose.bri.hp.com> In-Reply-To: <200010260037.e9Q0bGS54307@grumpy.dyndns.org>; from dkelly@hiwaay.net on Wed, Oct 25, 2000 at 07:37:16PM -0500 References: <sos@freebsd.dk> <200010260037.e9Q0bGS54307@grumpy.dyndns.org>
next in thread | previous in thread | raw e-mail | index | archive | help
On Wed, Oct 25, 2000 at 07:37:16PM -0500, David Kelly wrote:
> Soren Schmidt writes:
> > The only drives I can recommed are IBM's, but beware the new DTLA series
> > are so fast that some old controllers can have problems with them.
<snip>
> Have seen bonnie report over 22 MB/sec on that drive and system. Then
> again this second one is almost sad enough to be funny. OTOH it is
> reliable:
ad0: 73308MB <IBM-DTLA-307075> [148945/16/63] at ata0-master using UDMA33
This is the new version, bonnie says :
-------Sequential Output-------- ---Sequential Input-- --Random--
-Per Char- --Block--- -Rewrite-- -Per Char- --Block--- --Seeks---
Machine MB K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU /sec %CPU
100 11187 81.0 20539 42.0 2821 10.5 7282 77.5 28337 63.3 94.9 2.1
Sadly it's bit of a junk old box that has this drive in, But these
newer IBM drives are excellent, on a 3xRAM size file, iozone tells me:
14064030 bytes/second for writing the file
25977624 bytes/second for reading the file
It's also 75% full, so with that performance I'll be buying more.
(My experience with the previous IBM version (the 30GXP models) shows
that write performance is almost identical to read performance for the
first 75% of the drive, but drops to about half of the read
performance by the last few blocks of the disk.)
Also, I was under the impression that some of the large Maxtor drives
were for a while recently, just rebadged OEM IBM disks.
This disk is in a file server now though, so I can't test it out much
more (on a better machine perhaps). Over 100BaseT it outperforms the 4
controller striped 8x9Gb LVD disks on the K-Class it's attached to.
> Ditto. This is a system which has me reconsidering the wisdom of SCSI
> for my needs:
I was working with some 4Gb disk images (thank god FreeBSD handles
large files, not enough OS's do that!) last night, on the older 30Gb
IDE disk and although it is VERY fast, it doesn't cope as well as SCSI
as far as parellelism. While making a backup of the diskimage to the
same partition anything else I did on the machine was virtually
impossible, i.e. reading the image for ls off the disk and running it
took long enough for me to get bored and decide to make another cuppa.
Then again, I've not tried the above with SCSI, but I have a feeling it
would parellise better even if it performs slower for the same price.
> the purchase of a DEC 21140 based 10/100 PCI card for $12 and cured
I bought a dual 21140 (it's a J3514) for a HP K-Class for this project
with this IDE disk in a PC. Check out the price on that, for what's
effectively an ISA bus. We really are spoilt rotten with these PC
prices.
Steve Roome
P.S. These are all my own views, nothing to do with any employer I may
have had ever.
To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-stable" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20001030150026.B14579>