[conspire] DVI to VGA (was Re: CABAL this Saturday)
Rick Moen
rick at linuxmafia.com
Thu Jul 29 00:04:18 PDT 2010
Quoting Tony Godshall (tony at of.net):
> > Modern HDs are more reasonable than the antiques on my shelf, the best
> > of which are SCSI SCA 73 GB, 3.5" form factor. Those were fine drives
> > in their day, but that was a decade ago. I'm up for the higher
> > fragility of 2.5" SATA drives; that's what mirroring's for.
>
> Yeah, but 2.5" drives cost twice as much.
The thing is, I'm at a hardware-generational-change point. That's a
genteel way of saying the existing stuff's old, worn, low-capacity,
slow, large, and relatively expensive to operate. So, I have to get
_something_ soonish. Might as well get stuff that doesn't suck, has a
long economic service life, is operationally simple / no-hassle, and
at the sweet spot for technology versus cost.
Currently, in my view, that's 2.5" diametre SATA [/SAS] of about 2 TB
capacity. Desirable rotational speed / claimed seek times can be
debated, but personally I seldom do anything especially disk bound,
really don't care about those peformance deltas, and (mutatis mutandis)
prefer slower-spinning drives because they're cooler and draw less
power, hence less subject to heat stress.
Twice _what_? Whatever diddly little premium you're talking about,
amortise it over projected service life and deduct electricity-cost
savings, then add back the psychic reward of working with good tools.
> But I suppose you once you save enough power it pays back...
Going automatically for lowest initial acquisition cost is a fool's game.
That's not all that matters, nor even the most important thing, usually.
As I'm sure you already know.
http://linuxmafia.com/~rick/lexicon.html#moenslaw-bicycles
> Hey have you considered running the OS off compactflash and having
> physical hard drives power up only when the larger archival stuff is
> accessed? That would save even more power.
At the expense of introducing avoidable complications into my system.
Elegance is a virtue.
Using CF intelligently for system drives requires planning so as to put
there only what makes sense. Current mounts (some of which were for
now-obsolete reasons, and I'd not be doing them today):
:r /etc/fstab
# /etc/fstab: static file system information.
#
# <file system> <mount point> <type> <options> <dump> <pass>
proc /proc proc defaults 0 0
/dev/sda7 / ext3 defaults,errors=remount-ro 0 1
/dev/sda1 /boot ext2 defaults 0 2
/dev/md3 /home ext3 defaults 0 2
/dev/sda8 /recovery ext3 defaults 0 2
/dev/sda9 /usr ext2 nodev,ro 0 2
/dev/md4 /usr/local ext3 defaults 0 2
/dev/sda6 /var ext2 noatime,nodev,nosuid 0 2
/dev/md1 /var/lib ext3 nodev 0 2
/dev/md2 /var/spool ext3 defaults 0 2
/dev/md0 /var/www ext3 nodev,nosuid 0 2
/dev/sda5 none swap sw 0 0
/dev/sdb8 none swap sw 0 0
/dev/sdc8 none swap sw 0 0
/dev/hda /media/cdrom0 udf,iso9660 user,noauto 0 0
/dev/fd0 /media/floppy0 auto rw,user,noauto 0 0
Only limited parts of the server file tree make sense to put on media
with flash's wear characteristics. Note that this _is_ a server, such
that I really don't want to go 'noatime' in very many areas. (Stuff
breaks. Stuff isn't right.)
And as to /tmp, /var/tmp (be those tmpfs or not), and /var/lock, NO.
On _flash_, Tony? That makes no sense. Horrors. 'Minimize logging'?
Are you not clear on this being my _server_? No, it makes no sense to
minimise logging on a general-purpose Web / shell / ssh / rsync / SMTP /
ftp / mailing list server.
Basically, I'd find some way to use flash _if_ it came with the system
unit, but it's a hassle to deploy if limiting it only to parts of the
filesystem without frequent writes (which of course includes atime).
A key aim of home server design is minimal hassle. Like, for example,
md-driver RAID1 mirroring, which is a thing of beauty and power for busy
sysadmins wanting home servers that are functional, reliable, low-cost,
and hassle-free. I'd have to md-mirror a _pair_ of flash devices to get
that same degree of win. _And_ have an md-mirror of HDs.
Overengineering. Feh.
More information about the conspire
mailing list