Bacula-users

Re: [Bacula-users] Advice needed on Linux backup strategy to LTO-4 tape

2009-08-12 19:36:53
Subject: Re: [Bacula-users] Advice needed on Linux backup strategy to LTO-4 tape
From: Arno Lehmann <al AT its-lehmann DOT de>
To: bacula-users AT lists.sourceforge DOT net
Date: Thu, 13 Aug 2009 01:31:37 +0200
Hello,

13.08.2009 00:19, rorycl wrote:
> I'm going to cross-post this text on the Amanda and Bacula lists.
> Apologies in advance if you see this twice.

Great, let's start a cross-list flame war ;-)

Welcome here anyway ;-)

> Our company is about to provide centralised backups for several pools of
> backup data of between 1 and 15TB in size. Each pool changes daily but
> backups to tape will only occur once a month for each pool.

Meaning that the monthly backups can get rather big, even if you don't 
do full backups each time... or are you planning to do only fulls? 
Then I hope you're not forced to use a slow and unstable WAN link...

> The backup tape format is to be LT04 and we have a second-hand Dell
> PowerVault 124T 16 tape autoloader to work with currently. Backup from a
> pool may be taken off a Linux LVM (or hopefully soon a BTRFS) snapshot
> ensuring that the source data does not change during the backup process.

A good start.

> We have the possibility of pre-preparing backup or compressed images if
> this is advisable.

That's not needed with Bacula - creating and destroying the LVM 
snapshots can be integrated into the jobs (as well as ignoring the 
directory you mount the snapshot to, ensuring only the original path 
names are stored).

> An important aspect of the system is that the tapes should be readable
> for 12 years,

So I hope you've got the proper storage conditions.

> by other parties if necessary.

No Problem.

> From this point of view we
> like the idea of providing a CD with each tape set of the software
> needed to extract the contents,

Simple with Bacula - you just need the bls and bextract programs. I 
would advise to distribute those as source file - in 12 years, you 
will have difficulties finding a system where today's executables will 
be of any use.

Also consider that finding the needed hardware in 12 years might be 
difficult, so you'll have to set up a routine to verify the hard- and 
software needed to access the data is available.

Bacula's tape format is, obviously, open source, so it's not a problem 
per se to ensure you can read it, but just to feel comfortable I would 
create a rescue media each year which runs Bacula (or at least bls / 
bextract) on the then-current hardware, and has the needed drivers to 
access whatever drives are then available to read your tapes.

> together with a listing of the enclosed
> files in a UTF8 text file.

Creating that is just a simple SQL query, which actually already is 
distributed with Bacula - have a look at the query command in bconsole:

> *que
> Automatically selected Catalog: BaculaCat
> Using Catalog "BaculaCat"
> Available queries:
...
>     12: List Files for a selected JobId
...
> Choose a query (1-16): 12
> Enter JobId: 23456
> +------------------------------------------------+-------------------------+
> | Path                                           | Name                    |
> +------------------------------------------------+-------------------------+
> | /dev/                                          |                         |
> | /proc/                                         |                         |
> | /root/                                         | .asterisk_history       |
> | /root/                                         | .bash_history           |
> | /sys/                                          |                         |
> | /tmp/                                          |                         |

You can create that list automatically by a script which is run after 
each job.

> We will be required to audit each backup set
> by successfully extracting files from tape.

No problems expected.

> We are very familiar with working on the command-line in Linux,
> Postgresql and Python.

I don't think you will need much python for your tasks :-)

Creation (and ensuring it's useful) of a mini-linux distribution to 
read your tapes wouldn't be too much of a problem, too.

The biggest part of your work, when using Bacula, will be to set up 
and maintain the proper procedures and auditing, which is mostly 
writing text :-)

> As we have not run backup to tape on Linux before I would be very
> grateful to receive advice on what approach members of this list would
> take to meeting the above requirements.

Actually, I would start with a rather plain Bacula setup. Just make 
sure your tapes are never automatically reused, and ensure the files 
are pruned from the jobs in time, so your catalog doesn't grow 
excessively. Set up, test, and automate a build environment for your 
add-on media which has the tools and file list included.

Finally, glue it all together to make the add-on media creation fully 
automatic... and then starts the real fun - testing, maintaining, 
auditing.

Actually, I don't think that what you need is much of a challenge for 
Bacula and an experienced sysadmin :-)

Cheers,

Arno

> Many thanks,
> Rory
> 

-- 
Arno Lehmann
IT-Service Lehmann
Sandstr. 6, 49080 Osnabrück
www.its-lehmann.de

------------------------------------------------------------------------------
Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day 
trial. Simplify your report design, integration and deployment - and focus on 
what you do best, core application coding. Discover what's new with 
Crystal Reports now.  http://p.sf.net/sfu/bobj-july
_______________________________________________
Bacula-users mailing list
Bacula-users AT lists.sourceforge DOT net
https://lists.sourceforge.net/lists/listinfo/bacula-users