Archiving Projects-- TAPE vs. DVD

Ryan Simpkins plug at ryansimpkins.com
Fri Feb 10 09:58:09 MST 2006


On Fri, 10 Feb 2006, Ross Werner wrote:

> On Fri, 10 Feb 2006, "Gregory Hill" wrote:
> > I had a hard drive with a 3 year warranty that I used for probably 2
> > years and let sit on the shelf for 5 or so.  Picked it up one day,
> > plugged it in, and everything was still there.  Drive still worked
> > great.  Maybe that's abnormal, but I thought it was worth sharing.
> 
> I've never had problems with a hard drive that had been sitting unused for 
> years. If it's used constantly for years, I can see it getting old and 
> wearing out, but I've never seen a problem with a drive that hadn't been 
> used for years.
> 
> (It's probably worthwhile to note that the opposite has been true with 
> optical media; a signficant percentage of the writable optical media that 
> I've used that had been sitting around unused for years has gone at least 
> partially bad.)
> 
>  	~ Ross

I did some quick googling on the subject of tape/optical/hard drive 
storage. Here are a couple helpful articles I found:

(How long does tape last, really?)
http://searchstorage.techtarget.com/tip/1,289483,sid5_gci989561,00.html 

(Long-Term Usability of Optical Media)
http://palimpsest.stanford.edu/bytopic/electronic-records/electronic-storage-media/critiss.html

Sorry, but I couldn't find any articles on people using hard drives for
long term data storage. Maybe someone can. My suggestion is to stay away
from it as it is not (at this time) a technology really intended for this
purpose.

Now I can tell you what I do. I have 230 Linux servers that need to be
backed up every night. Long-term data retention is not a critical piece of
what we need, so I realize this probably won't fit the bill for you. I
need maybe a week of on-line backup at the most.

Right now we are using low cost storage arrays (RAID5) and rsync
"snapshots". We have tried both tape and bacula in the past. I have to say
that my experience with both of those technologies was frustrating to say
the least. I'm not saying these are bad technologies, they just didn't
work for us and what we needed.

The rsync solution was great. It was easy to write a fun little web-based
front end to the application which includes monitoring/paging. At a glance
I can see real-time information on how backups are moving along, estimated
completion time, etc. I wrote client software that makes mounting the
backup read-only over NFS a snap. The backup system will even detect newly
provisioned systems, auto-configure them, and start backing up. I also
wrote the ability to auto-recover from crashes/freezes/reboots. The code
was not that complex at all. It got done in a couple of weeks.

Right now I have about 24TB of on-line backup storage, that gives me about
5 nights worth of data - plus plenty of room to grow. I backup about
330,000,000 files. The cost was much lower than other technologies we
could find, and it works reliably.

Here is an article on rsync snapshots:
http://www.mikerubel.org/computers/rsync_snapshots/

All in all this was a totally custom solution for us and our needs. You 
probably have totally different needs, and are welcome to ignore this 
entire e-mail.

-Ryan




More information about the PLUG mailing list