Subject: Re: cluster waste?
To: NetBSD Users <netbsd-users@NetBSD.org>
From: Chuck Yerkes <chuck+nbsd@2004.snew.com>
List: netbsd-users
Date: 03/24/2004 01:23:53
Quoting Matthias Scheler (tron@zhadum.de):
> On Thu, Mar 18, 2004 at 09:04:07AM -0500, Steven M. Bellovin wrote:
> > Has anyone revalidated those numbers for modern disks? When FFS first
> > came out, there was a lot of talk about rotational optimizations,
> > staying within the cylinder, etc. Nowadays, of course, that's all
> > mythical.
>
> And 10% of a current disk are more than 10GB. That seems like a *lot* of
> space to waste. The minimum free space threshold should probably be
> determined automatically during "newfs" based on the filesystem size.
Mr McKusick mentioned the same thing when disks were 2 GB or so.
Along the lines of "200 *MEGA* bytes" free? That's is larger than the
the disks [they] had when [he] wrote 10% and it's silly now.
While it ALWAYS depends on the files (storing mostly 500MB - 1GB files on
a 120GB drive is about the same when it's an order of magnitude smaller),
I usually run 1-2% free on 60GB+ drives. My files are about the same
as they were 10 years ago, except for those mp3s on a separate partition.
(yes, I own all the CDs, I just haven't unpacked them since last move )
It leaves enough space to move several files for optimizations.
FreeBSD, I believe, defaults to 5% (don't quote me on that).
With real RAID boxes (external, hard RAID attached via fiber or SCSI),
we usually have a decent NVRAM between us and the actual disks, so
cylinders is all myth. But with real disks, SCSI or not, there ARE
some considerations for placement.