[lug] LVM and disk failure

[Sean Reifschneider]

On Sun, Jan 08, 2006 at 12:00:29PM -0700, Daniel Webb wrote:
>For my 1000-disk example, I'd be a little worried even with RAID.  RAID

I don't understand your point...  The controller can go haywire in a single
drive system just as easily with exactly the same results...

LVM is all about convenience.  It'd better be, because it takes a hefty
performance penalty.  I run it in 2 places:

   Our virtual dedicated hosting machines, where we need to create an
   arbitrary number of 1.5, 3, and 6GB partitions, plus 50, 100, and 200MB
   swap partitions.  This is really the only solution on a system with 15
   or 30 of these partitions...

   My home backup server where I want to be able to add more discs as I
   need more capacity, without really caring about peak performance or
   losing the vast majority of data.  All I really care about on the backup
   server is the backup control and config files, the rest of the files on
   the system is data that we have on other machines...  As long as I
   detect it when it fails and not when I need to recover, I'm ok.

For these uses, LVM is great.

I have, in the past, run LVM volumes totalling around 150 drives on 3
systems, and with RAID-1 we never had a data-loss, even though we did have
drives fail about once every 6 weeks.  They were hot swap so it was no
problem.

Thanks,
Sean
-- 
 It often shows a fine command of a language to say nothing.
Sean Reifschneider, Member of Technical Staff <jafo at tummy.com>
tummy.com, ltd. - Linux Consulting since 1995: Ask me about High Availability
      Back off man. I'm a scientist.   http://HackingSociety.org/