[ale] Hard Drive Failures

[JD]

I'm with these guys.

I was looking at the backup options again and there are many more now that do
all the right things and seem to do them really well, at least if you look in
Synaptic (software center seems to be a joke to me).

I am still using rdiff-backup, but some recent failures that I've been unable to
resolve on a desktop have me looking for a alternative. It has been working
almost perfectly on 10+ servers almost 5 yrs besides that.

If you are using rsync alone and getting mirrors, you are missing out on some of
the best parts of backups - versions. It is possible to use hardlinks and rsync
to get versioned backups, but those are not as efficient as rdiff-backup on
storage. Just sayin.

For every 1,000 backups, are you certain that you can restore at 3am after a
night of drinking? If not, your restore process is too complex.  If it takes a
month to get your data back, is that acceptable too? Online backups are not all
they seem to be for large amounts of data.

BTW, I was about to get sucked into the other thread too.  I've found my happy
place and will not post there.  Thanks David for keeping me out of that!

-jd

On 02/21/2013 05:14 PM, Jim Kinney wrote:
> +1e1M
> Don't worry about drive failure. Just be competent at restores from solid
> backups. Bit rot on tape is unmeasurably small compared to drive failure rates.
> If it matters, back it up. If you're not backing it up, it doesn't matter.
> 
> On Feb 21, 2013 5:05 PM, "David Tomaschik" <david at systemoverlord.com
> <mailto:david at systemoverlord.com>> wrote:
> 
>     So, I was going to let myself be sucked into Ron and dev null's debate on
>     the other thread, but decided better of it.  While I am no Steve Gibson fan
>     (in fact, I believe his products and advice are only slightly better than
>     snake oil) I'm going to steer clear of that and head for the central topic:
>     there are a *lot* of threads on this mailing list about drives failing; and
>     no, I'm not suggesting there's an epidemic of drive deaths, just that people
>     are approaching it the wrong way.
> 
>     If you care (aside from replacement time and cost) about a single hard drive
>     (rotational or SSD) failing: you are doing it wrong.  Plain and simple.
> 
>     Backups are the solution to hard drive failures, not ddrescue, spinrite, or
>     snake oil.  Pick any one hard drive in my place and destroy it.  Other than
>     buying and installing a new hard drive, I won't lose much sleep over that.
>      I might lose a little bit of data, but not enough that I'll be struggling
>     with ddrescue over it.
> 
>     JD and I once did a presentation at ALE-NW about backup solutions.  There
>     are many out there: commercial and open source, local and remote, network or
>     disk-to-disk, etc.  Pick one (or be like me and pick several) and use them.
>      Then enjoy the zen of treating hard drives as the commodities they are, as
>     opposed to the heart of your electronic world.
> 
>     (FYI, this all comes from a couple of painful experiences before I
>     was enlightened to the path of the backup.)
> 
>     -- 
>     David Tomaschik
>     OpenPGP: 0x5DEA789B
>     http://systemoverlord.com