[Dirvish] Replacing Tape...

[Noel Kelly]

Hi All,

We have hatched a plan to replace our 500Gb tape backup with dirvish.  
We currently use dirvish to backup about 7 file stores nightly to a 
1.2Tb NAS and it has been very successful for the last 3-4 years.

The tape backup has become too slow and too full.

So taking advantage of the huge sizes of HDD we have put together a plan 
which goes something like this:

First Monday:
HDD1 is inserted into eSATA hard drive caddy attached to central backup 
server.
Dirvish images initialised to HDD1.

Tuesday-Sunday:  Dirvish runs each night creating images on HDD1.

Monday:
index.gz (or log.gz) for each image is imported into an SQL database 
(see below) with heading metadata about HDD# etc.
Dirvish directory copied to backup server.
Dirvish directory copied to HDD2.
Dirvish images initialised to HDD2.
HDD1 removed offsite.

Repeat until HDD1 comes round again:

Monday:
Run dirvish-expire against HDD1 to purge all but the last images on HDD1.
(Work out a way to flag the database to show the backups have been purged)

We still need to work out how to design the SQL database to not 
duplicate every single file backed up etc.  The idea is to have a simple 
web interface to search for a regex to show the HDD instances containing 
that file.

Also would probably need some mechanism to check for available disk 
space and perhaps purge an image using the dirvsih pre-script before 
doing the backup etc.

Is there any obvious pitfalls in this strategy?  We thought if it works 
we can contribute the SQL schema and web pages back to the project.

Cheers
Noel