[clug] Story: Fijian Resort complex loses a single disk: business process stops for 1-2 days

[Eyal Lebedinsky]

Not exactly, but the idea is save.

I have two types of backups.

DR (Disaster Recovery) is a backup of the system (no user data, media etc.) which
can be used to rebuild a disk from scratch. It is taken on a long schedule, mostly
when there is a layout change or such.

Data backup includes all the data that I would rather not lose. I do this once a
month and use 5 disks in rotation (so not rewriting the same disk each time). Every
sixth month the backup disk goes "off site" for ever.

I should mention that Archival data (which does not change at all) is backed up
separately, off site, and not included in the same data rotation.

I do bring the "off site"s in every few years to read/rewrite them. A habit from
when I used tapes. Still not sure how long it is safe to leave data on a disk to
rot, but never had a read error (did have on tape though).

About RAID: it is not backup, but rather provides higher availability. Much
simpler to swap a RAID member on the fly than futz with the backup after a failure
(small or catastrophic). If I am away for a month, the array keeps going after
a disk failure until I get back and deal with it (RAID6 these days). This resilience
is nice to have. And if you have a family then your wellbeing may depend on providing
high availability :-)

cheers
	Eyal

On 07/25/14 23:18, Hal Ashburner wrote:
> Does anyone follow anything like the jwz prescription?
> http://www.jwz.org/doc/backups.html

-- 
Eyal Lebedinsky (eyal at eyal.emu.id.au)