System and data backup


I have a general question on what software do you use to backup the system and data.

Please advice.

Thanks in advance for your help!

Hi there!

We (company) generally do the backup stuff from under the machines as long as possible, so we snapshot the machine and letting the Backup solution (in this case Commvault/Veeam) handle the copying to an external share.

If it’s baremetal or not really accessible hypervisors it’s either once again one of these solutions or for workstations something like Déjà Dup Backup (uses Duplicati in the backend) works pretty well, file based instead of block based backups just take a whole lot longer for full backups.

I heard Duplicati itself should also work pretty well.
And I recently got in touch with Bacula the first time, which looks to be the go to open-source enterprise backup solution.

Hope that is a bit what you had in mind!

Restic backup is also very good.

This is mostly just for data, but I used rsnapshot throughout my work life, and currently still use it at home. I have successfully used a combination of snapshot and lsyncd to recover a system from scratch. The key here is that I used lsyncd to mirror the system in question and only recovered data from the backup that wasn’t recoverable for whatever reason. @lumarel solutions above are on point as well.

1 Like

I am very very old school. I’ve been doing Unix backups with dump for 30+ years. And it still works. So Sunday level0, Monday level1, Tuesday level2… For ext2/3/4 the dump command from the dump package; for xfs the xfsdump command from xfsdump package.

It’s not the most efficient in disk space usage but it has one big benefit; simplicity of restore.

See, a backup is useless if you can’t restore it. If you need to install some complicated software to read a database and provide a GUI… you’re making recovery hard.

The second part of a backup is “point in time recovery”. You might not notice a file is broken (corrupted, malwared, encrypted) for a few weeks. A tool that effectively just “offline mirrors” won’t let you go back 2 or more weeks; you get the latest copy or nothing.

For me, “dump” is a good compromise between “ease of restore” plus “a month recovery window” vs “efficiency of disk usage”.

IMHO, there are three datasets:

  1. OS files (from RPM packages)
  2. System config (including list of installed packages)
  3. User data

A configuration management system can maintain the config and OS files. If you have that set up, then you can “restore” system by re-install. That leaves only the user data to backup with aforementioned “copy” methods.

1 Like

Also, for consideration even if only as a tool in the backup box is timeshift which was recently added to epel after a period of absence.

Also, beware any blogs etc which do not cover a full recovery with the method discussed. I’ve wasted time on these finding out there are problems later. There are other long threads on backups on this forum though you will have to search hard sometimes to find what you’re looking for (not having a go guys).

We use bareos, because it can backup also databases and can be extend via C and python add-ons.

If you have to back up numerous workstations there is also BackupPC which I’ve used for years. It’s advantage is that it de-duplicates the files common to all machines via links after checksum analysis. I don’t know if it would be appropriate for large data sets though.

Actually, configuring rsnapshot correctly will get you backups that you can recover easily, including a single file from a particular day or hour. On top of that, it is super fast, because of the use of hard links in backup compares.

In the times when PA-RISC HP-UX systems were in use I worked with HP Omniback, later known as HP Data Protector and now Macrium Data Protecor. You can find 90-days trial here.
I liked this software since I could use it on HP-UX, Linux, Windows and Novel. All boxes in one backup environment called ‘Cell’.
Unfortunately it is far beyond my budget now… and looking something to do backups in the following way:
Httpd server backup:

  1. system backup - (ext4 filesystems) - full off-line backup done yearly or before/after system changes currently performed by Paragon Hard Disk Manager,
  2. data backup - every night for 12 weeks (/var/www/html directory and database export files copy).

I wonder how can I improve it?
Any suggestions please?

Just to join the party, NBER uses (I wrote) d2dbackup : d2dbackup - disk to disk backup program

I’m a cheapskate and lazy so using AMANDA for backup to a MHVTL virtual tape library that resides on a dedicated RAID 1 pair of 4TB drives. Been using AMANDA for not quite 30 years so no learning curve and it just works. MHVTL has been a “learning experience” but you can always substitute someone’s paid VTL if you have the budget. Paid VTLs generally allow you to create a physical tape backup for off-site, disaster recovery if you have that requirement.

For a last few days I’ve been testing Veeam Linux Agent Community Edition.
I wonder whether it is really free as they say…?

I have tested bare metal recovery from online backup… It worked fine…
Veeam by default creates filesystem snapshot while system is backed up (snapshot can be disabled).

The issue is that FREE edition has some features disabled - for example google drive integration… I was wondering whether it would be possible to mount google drive or after backups files are created sftp them or somehow replicate them there.

Has anyone used google drive for backups or maybe other storage service, if so, could you please advice on what your experience is.

Veeam will allow you to backup to a network drive.

Where is your local Google drive folder? I would try sharing the local Google drive folder over your network and using it as a Veeam backup target. Then the Google drive app should replicate it to Google’s servers for you.