Archival Best Practices

[jpbuse]

I'm looking for advice and all-around best practice(s) for archiving. I've read through the basic Retrospect guide on archiving but I'm looking for deeper info and some real-world examples to help guide me. I'm a bit new to archiving and while I have my idea of how to handle it, I'm assuming there are many others before me with much more knowledge they would like to pass down.

 

I want to archive old projects from our online storage to our offline (LTO3) storage. By "archive" I mean move to tape and delete from the source. After reading the Retro Tips docs, I plan to have 2 archive scripts. The first will archive w/o the optional move, and the 2nd will archive to different media w/ the optional move. This way I have two copies of the same archive on different tapes, one for onsite and the other for offsite.

 

My initial thinking was to have a server volume for my users called something like 'Archive', simple enough. I would train my users what to put there, how to name it, etc. This will get the data off the main data share and therefore out of my daily/weekly backup routine. Depending on the amount of data I would most likely commit to tape archive once every month. Once committed to tape, the projects would be removed from the online Archive share.

 

Although for this post I don't think tech specs matter too much, here is what I've got:

- All Mac OS X environment

- PowerMac G5 DP 1.8Ghz, 6.5GB RAM, 10.4.8 as dedicated Retro backup server

- Connected via FC is a 3.5TB xServe Raid and Exabyte LTO3 Magnum 224LC FC library

- Retrospect 6.1.126, DAV: 1.0.107, DUV: 6.1.7.101

 

Some issues/questions:

1. How will my users know what data is archived so they can retrieve it later down the road?

- I was thinking of using something like DiskTracker to catalog every single file in the archive and tag it with the corresponding tape IDs. A user could then tel me they need project X which is showing it is located in the June 07 archive on tape 1234.

 

2. If I commit a set of data to archive and it only uses 30% of a tape, is it ok to reuse that same tape next month to maximum the storage capacity of the tapes, or is that a no-no?

 

3. How would the process actually go, perhaps something like this?

a) move data on 'archive' share to date-stamped directory such as 2007.02

use DiskTracker to catalog all data in 2007.02

c) run archive script named 'ONSITE' w/o move using 2007.02 as source, once complete

d) run archive script named 'OFFSITE' w/ move using 2007.02 as source

e) move tapes to proper storage facility/location, etc.

 

4. Do I simply have 2 archive scripts that keep grabbing new tapes as needed... so in theory in a year I would have done 12 archives and it could have spanned across 10-20 LTO3 tapes. Is that ok, or is it best to treat each archive run as its own?