I am using Bacula Community 9.4.4 for backup of a server.

It has two drives with about 2.8 TB of information on each drive.

My backup system has ~28 TB of disk storage in a single partition using ZFS
0.8.5 with compression and deduplication turned on. I get about 1.3 - 1.4x
compression so it is equivalent to 36 - 39 TB of storage.

Bacula deduplication is not turned on.

I have backup jobs for each drive with daily incremental, weekly
differential, and monthly full backups.

1. The full backups are about 2.8 TB with 12,000,000 - 20,000,000 files and
each run for 2 - 3 days. This is a problem since it means that we can lose
up to three days of files if someone has a problem.
I am looking for suggestions about how to configure my backup schedule to
avoid this.

I can only get 3 - 4 months of backups out of this system. I set things so
pruning keeps enough volumes empty for the backups to finish.

Each volume has > 1 TB of files that do not change much such as cad tools.
These are not as critical since they can be downloaded and installed again.

Each volume also has about 1 TB of user files which are more critical
although most of them are not changing.

I am splitting the backups into separate jobs for the CAD tools and the
user data.

I am looking for ways to optimize this storage.

2. Any suggestions for backup strategies?

3. I would try Virtual backups.
Can I just run virtual full backups periodically after the first full
backup? I already have full backups so I could just schedule virtual full
backups instead of full ones.
Do I need separate media types or devices for Virtual Full backups? My ZFS
partition is already at ~ 80 % so I do not have much room.

Are Virtual full backups sort of like pruning so that my oldest backup data
is from when the last VirtualFull took place?

I looked at the following documents, alhtough they may be more for the
Enterprise version than the community one.

https://www.baculasystems.com/corporate-data-backup-software-solutions/bacula-enterprise-data-backup-software/features/synthetic-backup/

https://www.baculasystems.com/ml/pvf3.svg#PVF-title

4. I would be interested to try turning on deduplication. I looked at:

https://blog.bacula.org/whitepapers/DedupVolumes.pdf

I am worried about the statement below in the guide:

You must take particular attention to define a unique Media Type for
devices that are Aligned as well as for each Virtual Autochanger that uses
a different Archive Device directory. If you use the same Media Type for an
Aligned device type that you use for a normal disk Volume, you run the risk
that you will have data corruption on disk Volumes that are used on Aligned
and non-Aligned devices.

In my bacula-sd.conf file I had already specified most of the settings for
deduplication. Note that even though the device is called Aligned-Disk it
was not specified as aligned disk.

The problem again is my ZFS partition is already mostly filled.

If I make a separate media type then can I point it at the same disk
partition or do I have to use a separate partition so that new disk volumes
will be created just for the aligned storage.

My concern with deduplication is that it uses a lot of ram so I really need
to test it with my largest backups to see how it goes. I have seen numbers
like 5.5 GB of Ram/TB of storage. For my system that would be > 150 GB ram
and I only have 48 GB. It is probably cheaper to get a new backup system
with more drives than to start adding tons of RAM.

Device {
  Name = Aligned-Disk
#   Plugin Directory = /opt/bacula/plugins
#  Device Type = Aligned  # Must be aligned
  Media Type = File1
  Archive Device = /mpool1/bacula    # Or /mnt/ddumbfs.mnt if ddumfs
mounting point.
  LabelMedia = yes;
  Random Access = Yes;
  AutomaticMount = yes;
  RemovableMedia = no;
  AlwaysOpen = no;
  Maximum Concurrent Jobs = 1    # Always 1 for Aligned
  Minimum Block Size=0K
  Maximum Block Size=128K
  File Alignment=128K
  Padding Size=512
  Minimum Aligned Size=4096
}

Device {
  Name = FileChgr1-Dev2
  Media Type = File1
  Archive Device = /mpool1/bacula
  LabelMedia = yes;                   # lets Bacula label unlabeled media
  Random Access = Yes;
  AutomaticMount = yes;               # when device opened, read it
  RemovableMedia = no;
  AlwaysOpen = no;
  Maximum Concurrent Jobs = 5
}

Thanks
_______________________________________________
Bacula-users mailing list
Bacula-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/bacula-users

Reply via email to