[nylug-talk] Slim home server for samba and subversion ( and possibly IMAP)
Bryan J. Smith
b.j.smith at ieee.org
Fri May 16 00:30:45 EDT 2008
On Thu, 2008-05-15 at 01:14 -0400, Chris Knadle wrote:
> I've installed several *NIX OS's via BOOTP/TFTP including PXE, but I'm not
> sure what you mean by "legacy" BOOTP/TFTP, and I've never attempted to load
> Windows via BOOTP. Just... somehow never came up.
Oh, sorry, I should have provided some context.
Too many Windows admins I know are of the type that only know "product
names," not "technology implementation." So most literally think
Microsoft not only invented the technologies it uses in its Remote
Installation Service (RIS), but are wholly ignorant that the UNIX world
has been using the exact protocols same for decades.
BOOTP (which DHCP is based on, including the RARP and the fields used
for boot parameters) and TFTP are virtually the same. In fact, at my
client, the limitations of Microsoft's own RIS solution is why we use
Anvin's PXELINUX, including various COM objects, for booting various
Windows WinPE ISOs and 3rd party images (for Windows).
So, my point was, when I start talking about such things, they can't
follow anything I'm saying. Eventually they stop me and go, "do you
know RIS?" And all I can answer is, "I know RIS, and all of its
limitations and reasons why enterprises don't use it, but another
solution -- often PXELINUX." Again, they don't follow at all. ;)
[ FYI, I keep my "Minesweeper Consultant and Solitaire Expert" (MCSE)
and related Microsoft paper non-sense "current" for a reason. Not
because I believe in it, but when you have more Microsoft credentials
than the engineers from a Microsoft Gold Partner, it tends to win you
more contracts. Especially when you're selling against Microsoft. ]
> Well, again, I use the above settings for drives that are
> only /occasionally/ used. Setting the drive to spin down after 15 mins of
> inactivity is pointless on the drive that contains the root (i.e. /)
> filesystem; it seems to always get some periodic activity and in my
> experience never spins down.
> The reason I bother to do this is that I have several drives in internal
> enclosures in desktop boxes that are used for backup purposes, so the only
> time they're used is during the backup. Thus most of the time the drive
> spins up in the morning, is never even mounted, spins down after 15 minutes,
> and is never spun up again for the rest of the day. If and when a backup is
> done the drive automatically spins up again when it's used, and spins down
> when it's done.
> I agree with you that for a drive that's sporatically used throughout the
> day setting it to spin down after 15 minutes is a bad idea, since it will
> pointlessly end up having the drive spin up + down again lots of times every
> day. But in my particular use case I think what I'm doing is fine. However
> as I mentioned, I don't know if it actually helps increase drive longevity or
> not. My understanding is that the main reason this is done for laptops isn't
> for drive longevity but rather for power savings, and being too aggressive in
> spinning down the disk is detrimental for the drive.
I'd make it more like 3 hours or so. But that's just me. You know your
habits better than I do.
It's just ideal if you power your commodity 3.5" drives on and off no
more than one time per day. There is also the power v. actually
spinning detail.
> When it comes to drives only being used 8 - 14 hours / day, well, I don't
> disagree, but I think there's just not much to be done about it. Obviously
> I'm not going to try to power down my remote servers for half of every day
> just so that the drives can be spun down. ;-) So as far as I'm concerned
> the length of time per day that I use a disk is arbitrary, and I'm just not
> going to worry about it -- it is whatever it is. I'm likewise probably not
> going avoid working on my laptop just because my "allocated" disk usage time
> for the day is up. :-P
That's why every vendor introduced the 24x7 "Enterprise" commodity disk
about 5 years ago. Originally the name was "Near-line" meaning they are
not spinning 24x7, but powered 24x7. They are fabbed the same way as
"consumer" commodity disks, right off the same assembly, but they test
to lower vibration and better thermals.
E.g.,
Hitachi 7K -> T7K (no idea what the "T" standards for)
Seagate Barracuda 7200.7-9 -> NL35 ("Near Line")
Seagate Barracuda 7200.9-11 -> Barracuda ES ("Entry (?) Server")
Western Digital Caviar SE -> Caviar RE ("Standard" v. "RAID" Edition)
--
Bryan J Smith Professional, Technical Annoyance
mailto:b.j.smith at ieee.org http://www.linkedin.com/in/bjsmith
-------------------------------------------------------------
Fission Power: An Inconvenient Solution
More information about the nylug-talk
mailing list