Multiple Storage Backends


I’m new in the managed backup scene. Until now we did a very poor backup job (each server has a backup server and a hard disk on it, where the backups are stored, no RAID, backup scripts run in cron jobs).

Now I’m searching for a better solution, and installed a ceph cluster.

We have lots of old hardware which we can use, the only downside is, we are limited to 1Gbit/s.

So, is it possible to have a restic management node, which is “just” managing the backups?

Ideally there would not be much Traffic through this node, just between the different storage nodes and the servers.

Thank you for your answers

Hey I’m also new to restic, but I saw this post which might have some interesting info for you: CERN is testing restic for their backups.

restic just writes its backups directly to the repository location. For ceph you’ll probably want to configure it in restic as a S3-like backend. In that case the backup should be sent directly to the storage nodes.

What functionality do you exactly mean by “managing”? Restic itself just handles the backup part, scheduling backup runs and so on requires additional scripts. You might want to take a look at the different restic setups presented in some of the forum threads.

Using cron to run your backup scripts isn’t necessarily a bad idea as long as you check that the restic runs completed without errors and setup monitoring to ensure that the backups are actually created. (and run backup checks regularly)

I just realised, restic isn’t a management software like i had it in mind. So you have to install it on every server and configure it there, to make backups to the desired backend/repositroy? So you don’t have a central location where you can see all the backups and manage them?

If your management node can read data from the servers, then restic doesn’t need to be installed on them. It can work over various backends, all listed in the documentation. By “install and configure”, keep in mind that restic is a single binary so you may find it easier to distribute than you might expect :slight_smile:

Correct, but your implementation of restic could do this for you. Each client system can backup to a single repo which is then checked/pruned by the management node itself, producing its own logging/reporting.