[nycbug-talk] off site backup

[alex at pilosoft.com]

On Sun, 30 Oct 2005, pete wright wrote:

> this raises an interesting question, for me at least.  I'va had bad
> tapes in the past that I have been able to recover from, minus some
> missing data - but I was still able to dd the archive off the tape for
> testing and eventual recovery.  how do people do this on a drive, when
> say after a year or two of shelf life the disk itself is not spinning up
> or some blocks get curropted?  this is pretty much why I still don't
> trust using an IDE HD based backup solution...heck I'm not even sure I'd
> trust a SCSI HD solution if it was affordable...
Disk drives fail in different ways. It is uncommon for drive not to spin 
up after 1-2 years of shelf life - it is the *power on hours* that usually 
kill you. Yes, there are certain failure modes that result in drive being 
completely inaccessible - motor burning out, failing bearings, broken head 
- but those are fixable (at about 1000$ cost) by a competent guy with a 
donor drive, a soldering gun and a screwdriver. ;)

Keep in mind that usually, with hard drives, it isn't hard to implement a 
RAID system where failure of one drive doesn't impact your entire system - 
much harder than tape, as you need >1 tape drive to make it happen.

Keep in mind that media for tape (say, ~40$ for LTO 200G) is about 1/4 of 
cost of similar hard drive media.

So, its really apples and oranges. For historic (say, you have to do once 
a week backup and store it offsite, with retention of 1 year) - it doesn't 
make sense to do hard drives. For near-line and on-line, you want to use 
hard drives.

-alex