Adam Goryachev wrote at about 00:14:53 +1000 on Thursday, April 28, 2011:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> Discussing the differences between using a 2 member RAID1 where one
> device is periodically removed and rotated offsite to a 3 member RAID1
> where one device is periodically removed and rotated offsite.
>
> On 27/04/11 22:47, Jeffrey J. Kosowsky wrote:
> > I still think that losing all 3 (which however unlikely is still
> > possible) is way, way, way, worse than potentially losing 1-2 out of 3
> > and still having a spare to recover (carefully) from. And my case can
> > occur if you lose a disk controller or if there is a transient or if
> > you do something stupid and overwrite the disk, etc...
>
> I'm not sure why you think you would lose 3 out of 3? Sure, with the 3
> member RAID1 setup, you need one additional drive compared to your 2
> device RAID1 setup.
>
> ie, 2 RAID1 devices are always installed, another 3 devices are rotated,
> with one offsite, one onsite, and the third in one of (transit, offsite,
> onsite).
Well, this is the first anyone mentioned about "another 3 devices are
rotated". Of course, if you are allowed to posit lots of other offline
backups then be definition you risk less even if all of your current
RAID devices fail. The OP however seemed to imply a *single* 3rd
device not 3 other devices...
> Thus, if your 'worst case' of the three member RAID1 going bad,
> then you still have the device you just failed out of the RAID1, as well
> as the final offsite RAID1 member.
Well if you 'failed' one device then you have essentially the case
that I am advocating for! i.e., leave one of the original copies in a
non-writable state so that you don't inadvertently lose everything.
That being said, removing the drive physically if possible is even
safer than just software failing it.
I think what people don't fully realize is that no matter how good
your software RAID is, a hardware problem can easily result in an
error propagating across all live RAID1 copies... as I discovered,
this is a very real and painful case...
> Thus you need to lose both RAID1
> devices during a resync before you need to rely on the device you just
> removed, or finally the offsite copy.
Which is *precisely* what I was proposing except that in addition to
failing the device, I suggested removing it physically for extra
security (again assuming you don't have 3 other backup devices
rotating around offsite :P)
>
> Alternatively, with a max 2 member RAID1 device, you only need to lose
> one drive during a resync to have the same result.
No one talked about that case...
> BTW interface failure, controller failure, etc are all of equal risk in
> either case, so IMHO, that can be ignored as far as determining the
> cost/benefit of these two particular cases (but of course, should also
> be considered somewhere else).
But if a device is physically removed and failed then it is
independently safe unless something physically happens on your site
that destroys both the removed disk and the active RAID1 disks.
------------------------------------------------------------------------------
WhatsUp Gold - Download Free Network Management Software
The most intuitive, comprehensive, and cost-effective network
management toolset available today. Delivers lowest initial
acquisition cost and overall TCO of any competing solution.
http://p.sf.net/sfu/whatsupgold-sd
_______________________________________________
BackupPC-users mailing list
BackupPC-users AT lists.sourceforge DOT net
List: https://lists.sourceforge.net/lists/listinfo/backuppc-users
Wiki: http://backuppc.wiki.sourceforge.net
Project: http://backuppc.sourceforge.net/
|
