From owner-freebsd-scsi  Fri Sep 22  1:32:18 2000
Delivered-To: freebsd-scsi@freebsd.org
Received: from smtp05.primenet.com (smtp05.primenet.com [206.165.6.135])
	by hub.freebsd.org (Postfix) with ESMTP
	id B16DC37B42C; Fri, 22 Sep 2000 01:32:13 -0700 (PDT)
Received: (from daemon@localhost)
	by smtp05.primenet.com (8.9.3/8.9.3) id BAA23297;
	Fri, 22 Sep 2000 01:32:30 -0700 (MST)
Received: from usr05.primenet.com(206.165.6.205)
 via SMTP by smtp05.primenet.com, id smtpdAAAmJaGDT; Fri Sep 22 01:32:24 2000
Received: (from tlambert@localhost)
	by usr05.primenet.com (8.8.5/8.8.5) id BAA09658;
	Fri, 22 Sep 2000 01:32:02 -0700 (MST)
From: Terry Lambert <tlambert@primenet.com>
Message-Id: <200009220832.BAA09658@usr05.primenet.com>
Subject: Re: disable write caching with softupdates?
To: peter@netplex.com.au (Peter Wemm)
Date: Fri, 22 Sep 2000 08:32:02 +0000 (GMT)
Cc: sos@freebsd.dk (Soren Schmidt),
	Stephen.Byan@quantum.com (Stephen Byan),
	mbendiks@eunet.no ('Marius Bendiksen'), fs@FreeBSD.ORG,
	sos@FreeBSD.ORG, freeBSD-scsi@FreeBSD.ORG
In-Reply-To: <200009220757.e8M7vtG46023@netplex.com.au> from "Peter Wemm" at Sep 22, 2000 12:57:55 AM
X-Mailer: ELM [version 2.5 PL2]
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Sender: owner-freebsd-scsi@FreeBSD.ORG
Precedence: bulk
X-Loop: FreeBSD.org

> > So, having the write cache there definitly is a win.
> 
> It is a win only if you do not value your data.  I would gladly turn it off.
> How do we do this right now?   (ie: completely off)
> 
> > I'll try this on TWO IBM DTLA drives with tags enabled and see what gives..
> 
> I'm curious to know if tagged queueing compensates for the loss incurred by
> disabling write caching.

For a multiple application system, the anser is "yes", so long as
there are enough applications that the average stall time for a
context switch equals or exceeds the latency induced by waiting
for the write to complete (standard queueing theory 8-)).

For a single user workstation, where there is generall only one
or two applications running, the answer would be "no".

For you to be able to see this, the appropriate test is probably
a "make world" (or a similar large multielement compile) with an
argument between "-j 8" and -j 12".

Note that you should probably _not_ use a memfs /tmp for this, at
the same time, since what you want to stress is the impact on
concurrency under a multiprogram load.


					Terry Lambert
					terry@lambert.org
---
Any opinions in this posting are my own and not those of my present
or previous employers.


To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-scsi" in the body of the message