Ralf Auer wrote:
> Hello everybody,
>
> I need a little help from someone more experienced with Bacula than I
> am.
>
> My cluster consists of ~40 clients, the total amount of data that has to
> be backed up is around 6 TB. To accomplish this I have two HP Ultrium
> 960 (400/800GB) and a sufficient number of tapes.
Sounds like a pair of LTO3 drives. Are they just a pair of standalone drives,
or do you have a library with a tape changer?
> I am planning to have a 'full' backup of all my clients every two or
> three weeks and 'incrementals' all other workdays.
You may also want to throw in some differential backups in there as well, to
a) cut down on the number of tapes that will be required to go through on a
restore, and b) reduce the impact of losing a tape.
> My plan is, to have about three computers per night in the 'full'
> backup, all others in the 'incremental'. The next night, the next three
> in the 'full' and so on, so that after ~3 weeks all clients have been
> fully backuped and we will start with the first ones again.
Sounds quite reasonable - I'm doing a similar rotation here.
> According to the manual, the best way to do so is to define two pools, a
> 'full' and an 'incremental' which also allows me to write to the two
> tape drives simultaneously (one for each pool), which is mandatory since
> backups have to be done quite quickly at my site.
While this may be a good idea in terms of managing which backups go to which
volumes, but it's by no means necessary, especially if you've got a tape
library. I have a two drive library here, and have four simultaneous backups
going to two tapes in the same pool at a time via spooling.
> In addition I have to run my backups on a 'daily tape rotation' basis.
> So, if I understand correctly, I would need another 5 pools (mo-fri).
> That makes 10 pools for mo/fr, full/inc.
That's one option if you want to force a particular tape to be used on a
particular day. The usual reason is you've got exactly enough tapes for one
cycle, and you want to just cycle through them.
> Now I have two questions.
> * Does this setup sound reasonable to you, or did I get something
> completely wrong? I would be happy, if there was an easier way...
It sounds reasonable as a first pass.
> * How do I know, how many tapes I have to assign to each pool, because
> the amount of data is by far not constant per day and client. I do not
> want to have unused tapes, only because they were needed once.
Simple answer - enough tapes to hold however much data you back up =)
> Let me give you an example.
> Let's assume, that usually one tape for the 'full' and one tape for the
> 'incremental' per night is sufficient. Then I would assign one tape to
> each pool. During one Tuesday night the 'full'-pool runs out of tape,
> because some user has produced a TB of data the last day. Then, I guess,
> I would have to add another volume to the 'full-tuesday' pool and finish
> the backup on Wednesday morning, right? But what happens, if all other
> Tuesdays one tape is sufficient again ? Do I have to put a date sticker
> on that once used 'extra' tape and check/erase that tape manually after
> the first tuesday run where it was not needed so that I could use it for
> another day's oversize backup?
>
> You see, the very same setup was very easy in Amanda I used up to now.
> The only thing I had to do was to create two identical setups and put
> 50% of my clients in the one (using tape1) and the other 50% in the
> otherone using tape2, tell Amanda how many tapes I have and that's it.
> All the full, incremental and daily rotation thing is handled fully
> automatically.
Okay, more seriously, 90% of the time, the best answer in Bacula is actually
not to try to control that tightly what data goes on what volume. In this
case, for example, (assuming that you have a library) a better way might be to
simply set a volume use duration period of 24 hours. That would mean that 24
hours after each volume is used, it would be marked in the catalog such that
no more backups would be placed onto it until it gets recycled, which in turn
is controlled by the retention periods.
Also note that with a library, spanning tapes is a pretty trivial event for
Bacula. Once it hits the end of a tape, it simply grabs another volume from
the appropriate pool and pick up right where it left off. This is a
particularly big advantage over Amanda if you've got a single fileset that is
larger than your volume size, as (the last time I looked, at least) Amanda
couldn't span tapes.
Another advantage for you by doing this would be that you can simply stuff
your library full of tapes, and put them into a scratch pool. Bacula will
then simply shuffle a tape out of the scratch pool into the full or
incremental pool as needed, so you won't have to worry about guessing how many
tapes are needed for a particular night's backups.
Volume management in Bacula is fundamentally different from Amanda, and very
flexible, but that flexibility brings with it quite a bit of complexity. I'd
recommend reading the relevant chapters of the manual at least twice before
settling on a plan.
--
Frank Sweetser fs at wpi.edu | For every problem, there is a solution that
WPI Network Engineer | is simple, elegant, and wrong. - HL Mencken
GPG fingerprint = 6174 1257 129E 0D21 D8D4 E8A3 8E39 29E3 E2E8 8CEC
-------------------------------------------------------------------------
This SF.net email is sponsored by: Splunk Inc.
Still grepping through log files to find problems? Stop.
Now Search log events and configuration files using AJAX and a browser.
Download your FREE copy of Splunk now >> http://get.splunk.com/
_______________________________________________
Bacula-users mailing list
Bacula-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/bacula-users
