I have a script that moves backup disk volumes in a particular directory that are older than a certain number of days to Glacier; the volumes in that directory get populated by migration jobs from the primary pool.  

I'm skeptical about doing too much more for a couple of reasons. First, because I've found that I'm more comfortable dealing with Bacula's functionality from arm's length. I haven't had much luck, personally, with things like vchanger.  I'm a bit wary about trying to automate in a fashion that is too tied up in how things are currently implemented.

Second, I looked at my own use cases for recovery scenarios, and felt that locally-cached backup sets were a better fit for me, given the constraints of bandwidth and costs. I did some calculations on the cost of recovering significant data from Glacier, and depending on your time-to-restore requirements it can be _really_ expensive.  I'm in the process of adding code to my scripts to lengthen the recovery request submissions, but if there is a need for many GB of data from Glacier _now_ it might cost several hundred dollars.

I really think of Glacier as "cold" storage - secondary, in my case to locally stored volumes. It's great for that, but again, you really want to work out the costs before committing to using it with any expectations for actual restore processing. 

N. B.  one thing I have not looked at is using one of the available S3 mountable filesystem implementations as primary storage and setting up a process to move data from S3 to Glacier using the AWS life cycle tools,  if I were starting from scratch today, I'd look there first. 

Hope this helps, feel free to reply directly to me if there is anything I can amplify. 

_KMP

On Thursday, May 2, 2013, Ken Mandelberg wrote:

The last time I asked about this there was skepticism about using Amazon
Glacier for cloud storage of backups generated by Bacula.

I still have some interest in this particularly for off site backups for
disaster recovery at least for occasional snapshots.

Has anyone actually had success using Glacier with Bacula?

I have done some minimal tests using

https://github.com/uskudnik/amazon-glacier-cmd-interface/blob/master/doc/Scripting.rst

It's not very usable for daily backups as is. The backup volumes get
appended to until they reach their maximum and would be sent repeatedly
until they hit the max and a new volume is created. If I got around this
by some how forcing a new volume each time it would result in small
files being sent to Glacier, which is not desirable.

I suppose I could decouple the Glacier send from the Bacula director and
just send finished volumes when a cron sees that a new one has been created.

At any rate, just wondering if anyone is using Bacula with Glacier and how.


------------------------------------------------------------------------------
Get 100% visibility into Java/.NET code with AppDynamics Lite
It's a free troubleshooting tool designed for production
Get down to code-level detail for bottlenecks, with <2% overhead.
Download for free and get started troubleshooting in minutes.
http://p.sf.net/sfu/appdyn_d2d_ap2
_______________________________________________
Bacula-users mailing list
Bacula-users AT lists.sourceforge DOT net
https://lists.sourceforge.net/lists/listinfo/bacula-users

------------------------------------------------------------------------------
Get 100% visibility into Java/.NET code with AppDynamics Lite
It's a free troubleshooting tool designed for production
Get down to code-level detail for bottlenecks, with <2% overhead.
Download for free and get started troubleshooting in minutes.
http://p.sf.net/sfu/appdyn_d2d_ap2

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