[CentOS] Is Amanda "vaulting" what I need for archiving data?

Wed Jan 11 23:48:01 UTC 2012
Devin Reade <gdr at gno.org>

--On Wednesday, January 11, 2012 03:40:20 PM -0500 Alan McKay
<alan.mckay at gmail.com> wrote:

> Well, the scientists are talking longer than 7 years so HDs just are not
> going to cut it

Regarding the use of hard drives, you might want to have a look at this:
<http://www.lockss.org/locksswiki/files/ISandT2008.pdf>

At any rate, if you're concerned with archiving beyond seven years,
you probably also need to add another dimension to your archival problem.
In the professional archival/library industry they're quite aware of
having to maintain information longer than the lifetime of any of:

       - the individual media that it is stored on (eg: "the tape got
         too old and is now throwing errors")

       - the media type (eg: "my 10 year backups have been stored
         on ExaByte tapes in a humidity/temperature controlled 
	 vault, but I can't find a working ExaByte tape drive 
	 anymore", or "does anyone have a drive for my 8-inch
	 floppy?  How about a computer that will talk to the drive?")

       - the data format (eg: "the document I need is in AppleWriter
         format.  I was able to retrieve it from backups, and the
	 previously recorded checksums match, but I can't find a
         program that will read it!")

For long term storage, you may need to be able to not just put stuff
away, but also have a policy (and the resources!) to periodically
migrate data to newer media & formats.  This can get expensive in
time and money of course; your stakeholders may need to weigh in
again periodically to evaluate the value of the data vs the cost
of migration.

I'm sure that there are some archivist-related mailing lists 
out there that can better explain the depth of the horror.

Depending on the value of the data, you may also need to look at multiple
copies.  And then there's disaster recovery ...

Just because you didn't have enough problems already ...

(BTW, the main site <http://www.lockss.org> about LOCKSS looks 
interesting from an acedemic point of view, albiet not relevent
here.)

Devin