[Bacula-users] non standard configuration design

Hi all,

I'm new to bacula and thus have a few questions before trusting it with all my data.

Here is my configuration and the status of the system:
OS: GNU/Linux Debian stable (squeeze)
Bacula version: 5.0.2
Database: MySql 5.1.63
Hardware: DELL TL1200 (Single LTO-5 drive with autochanger of 24 slots).
    drive: IBM ULT3580-HH5
    autochanger: IBM 3573-TL
    Bar code reader present
Hardware seems to work fine
Tests run: btape test, fill, autochanger
All daemons are configured on the same server.

My main problem is to understand and determine the proper backup scheme as I am in a non standard situation.
My main issue being that the space to be used on tapes will be much larger thatn the space on hard drives.
The space available on hard drives is ~ 10TB. The space of archive estimated to 100TB/year, i.e. ~200 tapes for the retention period planned.
The fileSets are as follows:
    - git directory: standard backup
            - job type: daily incremental and weekly full or similar
            - pools: 'daily' and 'weekly' tape pools.

    - system files: standard backup
            - same as above but in a different job to simplify recovery.

so far it's simple, it becomes more tricky now:
    - rawData directory:
            - schedule: to be archived as soon as produced (trigger can be the absence of new data being saved to the folder ?)
            - the job name should preferably take the name of the subfolder (new dataset) in the folder specified in the fileset resource
            - size (here comes the problem): (~1TB every 2 days)
            - pool: 'data' tape pool (tapes will be replaced and added every 2 weeks or so).
            - NOTE: This folder should be emptied by a 'runafterjob' directive, as the folder cannot contain more than 1 or 2 datasets.

    - processedData directory:
            - schedule: as soon as produced (trigger can be manual by user), it seems specifiyng no schedule should work.
            - size: ~500GB produced every 2 days.
            - pool: One on tapes (pool 'data' as above or separate one), and a second one to NFS or SAMBA share (I don't know here if the file archive is appropriate or if a 'runAfterJob' rsync command could be better)
            - NOTE: This folder should be emptied by a 'runafterjob' directive, as the folder cannot contain more than 1 or 2 datasets.

My main concerns are:
    I want to keep the catalog for the rawData and processedData for at least 3years. Preferably without pruning anything and at the very least without pruning jobs.
    From what I understood in the manual, full backups always do pruning.
    The catalog should not increase above 2TB (or not much more). (I have 2TB of HDD that I can dedicate to that if needed).
    File size is generaly 1-3MB, fair amount of 100MB files, few of several 10/100GB.
    The catalog should be saved to tape only in the system job as it will grow in size. (I want only the part of the catalog that pertains to the jobs of a perticular tape on that tape)
    It might be interesting at some point to be able to force using a tape per job as the job size is almost that of a tape. However, I cannot be sure that the size will remain constant
    I do not know if I should create a big pool for the data pool or just a lot of small pools.
    I don't know what's the proper way to do that using bacula.

Thanks in advance and sorry for the long e-mail

Best regards,

Gnewbee

------------------------------------------------------------------------------
Live Security Virtual Conference
Exclusive live event will cover all the ways today's security and 
threat landscape has changed and how IT managers can respond. Discussions 
will include endpoint security, mobile security and the latest in malware 
threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/

_______________________________________________
Bacula-users mailing list
Bacula-users AT lists.sourceforge DOT net
https://lists.sourceforge.net/lists/listinfo/bacula-users