[8805] in linux-scsi channel archive
Re: What I learned at Tokyo Linux World.
daemon@ATHENA.MIT.EDU (Ricky Beam)
Sun May 14 13:23:23 2000
Date: Sun, 14 May 2000 13:20:52 -0400 (EDT)
From: Ricky Beam <jfbeam@bluetopia.net>
To: Ishikawa <ishikawa@yk.rim.or.jp>
Cc: LINUX SCSI <linux-scsi@vger.rutgers.edu>
In-Reply-To: <391C3271.FBF28A0B@yk.rim.or.jp>
Message-ID: <Pine.LNX.4.04.10005141312080.23096-100000@beaker>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
On Sat, 13 May 2000, Ishikawa wrote:
> - VA has chosen the SCSI boards they use very, very carefully
> based on the stability of software drivers, and
> then keep the driver in good condition by applying known
> patches, etc..
This is called "quality control"... while not new, few companies tend to
take the time now-a-days.
> - for really large high-end servers (meaning RAID, I think), VA
> has written a special driver for such controller cards that
> bypasses the SCSI subsystem totally, and the resulting
> driver makes the RAID looks as if it were a block device(!).
See also: DAC960
The Mylex RAID driver (DAC960) presents a block device to the kernel.
Internally, the controller is mostly SCSI-like. In theory, one could pass
SCSI commands through the driver to any SCSI device. There isn't any
support for this, however, and there's no intention to do so.
"For really large high-end servers" people will use hardware RAID controllers.
Software RAID is for the small-time operators. (Yes, I've used software
RAID in several places but I didn't care if the data was destroyed or
unavailable for days.)
--Ricky
-
To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
the body of a message to majordomo@vger.rutgers.edu
