Subject: Re: Useability of NFS over USB disks
To: None <firstname.lastname@example.org>
From: Jonathan A. Kollasch <email@example.com>
Date: 09/20/2006 20:18:47
Content-Type: text/plain; charset=us-ascii
On Tue, Sep 19, 2006 at 01:57:12PM +0200, firstname.lastname@example.org wrote:
> What about having a NFS server serving (external) USB2 disks ?
> I'm thinking about turning my Athlon64 barebone into a NFS server but
> there is only room for one disk. So I thought I might plug extra disk via
> USB2 is supposed to have a pretty nice speed nowadays but I heard about
> weird buffering/syncing process that makes USB disk not (quite)
> intensively useable.
> Anyone ever tried that ?
> Anyone knows if the USB driver on the motherboard (or on the disk rack)
> makes it better depending on the brand/version ?
I've gotten the impression that the ATA-umass bridge is the most
critical piece .
I have a brand-less PL-2507-based enclosure that reports that the
media has an extra sector, and then spews an error when I access
it. Other than that it seems to be ok, with faster-than-fast-ethernet
I also just got a PHR-100AC, a 1394/USB2 enclosure. On a 1GHz PIII
OptiPlex GX110 w/ a NEC-based USB2 board, I can get a whopping 15MB/s
via USB2 (raw device read rate). Via a NEC-based Firewire card in the
same box, I can get 38MB/s on the same test. On my 1.3GHz Celeron-
Dothan laptop (Compaq M2005US), FW performance is the same, yet USB
performance increases by 10MB/s. The whole thing seems to be of=20
a higher quality than the cheaper one.
I'm mostly using the brand-less one for random stuff w/ a 20GB drive.
I've even done a build.sh release on a LFS file system on this
The Macally is holding a 160GB drive that I do over-the-network
backups to, so the speed is limited to that of Fast Ethernet
(or worse yet, the performance of the file server's CPU) anyway.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.2 (NetBSD)
-----END PGP SIGNATURE-----