[prev in list] [next in list] [prev in thread] [next in thread] 

List:       toasters
Subject:    RE: Vol/qtree/dir/file snapshots (was Re: DataONTAP 6.2) ORACLE D
From:       Joe Luchtenberg <joe.luchtenberg () data-line ! com>
Date:       2002-03-29 16:50:29
[Download RAW message or body]

I think Chris' point has more merit than you give credit, Quentin.  There
are many issues to consider, including segregating the root volume, parity
disks, costs of additional shelves/cabinetry/head units, etc.  Granted the
cost of disk capacity does not scale linearly.  However, if you're not using
the additional capacity in a volume (30GB database on a volume of 5-6 72GB
drives, as in Chris' scenario) - and there may be very good reasons for this
- then big drives are definitely not the most cost-effective way to scale
capacity.
Qtree-level SnapMirror and/or file-level SnapRestore may solve some of these
issues related to capacity management, allowing you to put more databases on
the same volume.  Even still, it would be nice to have an internal 2x9GB
volume (maybe even with solid state disks!) to serve as root, rather than
"wasting" 72GB drives and shelf slots.

Just musing...

Joe


Joe Luchtenberg
Dataline, Inc.

Email:  joe.luchtenberg@data-line.com
Phone:  757-457-0504 (direct line)
        757-858-0600 (front desk)
        757-285-1223 (mobile)
Fax:    757-858-0606

Please visit us at www.data-line.com



> -----Original Message-----
> From: Quentin Fennessy [mailto:quentin.fennessy@amd.com]
> Sent: Friday, March 29, 2002 8:29 AM
> To: Chris Lamb
> Cc: Melvin, Grant; 'Noll, Kevin'; 'Toasters (E-mail)'
> Subject: Re: Vol/qtree/dir/file snapshots (was Re: DataONTAP 
> 6.2) ORACLE
> DATABASES
> 
> 
> 
> Chris asked:
> > Of course, one just begs to ask:  WHY is it that we 
> continue this absurd
> > notion that bigger is always better?  Who the hell needs a 
> 72GB BOOT DISK?
> > Sure, maybe when Windows(tm) 2004 comes out... :-)
> 
> This is a detail that is unnecessary to sweat.  Large drives 
> don't have to be
> full. The cost of capacity is dropping -- at a given time in 
> product life, a
> 72GB disk is not 8x the cost of a 9GB drive. It may well be 
> 1x the price,
> this 1/8x the cost!
> 
> Leave the disk space empty if its use is not appropriate.
> 
> If you need extra spindles for performance, buy them. Don't
> get caught in the raw capacity trap. 
> 
> I cannot imagine that it is economical to continue to sell
> relatively smaller disks, when for the same outlay the
> disk manufacturers can provide more capacity. The factories
> can provide X disks (or disk platters) per week. They will
> optimize appropriately. And they would go out of business
> if they did not, because the competition will!
> 
> As a system administrator we (you and I) all have to exercise 
> judgement on
> how to best manage these resources. If your boss or your 
> customers override
> your judgement then something should change. But it won't be the
> manufacturing economics of disk drives!
> 
> 
> > Yes, it is a fundamental law of computing that files and 
> databases will
> > always expand to fill the available disk space, but why is 
> it that you
> > can't even buy a 9GB drive anymore?  A 20-30GB database, even if you
> > double or triple it in size, is still going to fit on *one* 
> shelf with
> > 9- or 18GB drives.
> 
> > If I had a 30GB database, I'm sure my DBA would agree - 
> striping 5-6 9's
> > together makes way, way more sense than throwing piles of 
> higher-cost 36's
> > or 72's at it, only to waste massive amounts of space in an 
> effort to keep
> > performance up.  Of course, our database grows at over 30 
> million rows per
> > week, so adding 36's makes sense for us. :-)  Still!  I 
> echo the call for
> > using appropriately-sized drives where it makes sense.  :-)
> 
> > All of my production servers have ridiculously huge boot 
> drives because
> > that's the only way Sun (or anyone else) sells them these 
> days.  So I
> > mirror / & /var, interleave swap, and even after sizing 
> things to some
> > absurd degree, there's usually a chunk of 8-10GB of free 
> space - 'cuz all
> > the good stuff lives on the filers.  Having that local disk 
> space there
> > just means that someone is going to want to USE it, and 
> that's bad - it
> > means I'd have to actually back up those machines, and I 
> *hate* backups.
> > (In the exceedingly rare case where both drives of a 
> mirrored boot volume
> > are corrupted or destroyed, it's always going to be faster 
> to re-Jumpstart
> > and cfengine the machine from scratch than to restore from 
> tape anyway.
> > Local data just messes up all that beautiful automation. :-)
> 
> > A long time ago I was going to recommend that filers come 
> with a pair of
> > internal boot drives - like a mirrored pair of 4GB or 9GB drives
> > _strictly_ for use as the boot volume, with some space for 
> logs, etc.
> > There's plenty of room inside the big filer heads for a 
> pair of drives.
> > But the move to Flash RAM cards in the newest machines is 
> even better -
> > fewer moving parts. :-)
> 
> --
> Quentin Fennessy   		Quentin.Fennessy@amd.com	
> 				Office: 512.602.3873
> 				Cell:   512.694.7489
> 
> 
> 

[prev in list] [next in list] [prev in thread] [next in thread]