Planning off-site backups

It’s been quite some time since I’ve set up my own server and NAS at home. Since then I’ve thought a bit about backups.

On the NAS I not only store mine and my partner’s personal files but also a media collection that we share: primarily music and movies. Because the media collection was already significantly bigger than the personal files when I set the whole thing up, I decided to set up backups for the personal files first and figure out how to back up the media files later on. And I guess “later on” is now, or at least rather soon.

The NAS has two 8TB HDDs set up in a ZFS mirror, making the usable space for the entire NAS 8TB. Currently I use a little less than 3TB of this storage, and because of the storage for the VM backups I plan on having an off-site backup which can hold 5TB of data.

Since I don’t have a friend who has a storage server I can just install a few hard drives in, I will have to find a service that will fulfil my needs instead. I will judge these services by cost, business location, and their ability to let me do incremental backups with ZFS by using zfs-send(8) and zfs-receive(8) because I use ZFS extensively.

Possible providers

These are the providers I know of, including the location of their business and prices for 5TB of storage:

  • Hetzner Storage Box (Germany): 12,97€
  • Backblaze B2 (USA): $34,75
  • rsync.net (USA): $60,00
  • zfs.rent (USA): ~$20 + price of HDDs
  • Hetzner Dedicated Server (Germany): ~60-65€

If you know of any other interesting providers don’t hesitate to reach out!

The first problem with these providers is: all of them except Hetzner are based in the USA, which is something I’d like to avoid. The second: wow, most of them are really expensive!
I’m not saying they should be cheaper, I don’t know anything about the cost of running a backup service, but I’d certainly like to not pay up to 4 times the amount of my running server costs just for backups!

The most interesting of these is zfs.rent where you don’t get assigned storage from their hardware but instead ship your own hardware to them. Your storage is then accessible from a fairly lightweight VM on which you can do your backups. Each drive you send them costs you $10/month. Having your own VM with your own storage means backups with ZFS are possible.
The biggest downside here is ironically part of their unique model and bigger upside: sending hardware their way. Since I live in Germany this would incur pretty big shipping costs, including taxes.

rsync.net is another interesting one I’ve heard about a lot. They offer ZFS storage with a 5TB minimum, which is great since I plan on this amount of storage anyway, but that also means that this service would cost me $60/month. Additionally, they’re also based in the USA.

Backblaze is a very popular backup solution. I don’t think they have any special features or such, they certainly don’t let you access any ZFS pools, except that they’re fairly cheap. 5TB would cost me around $35/month, which is still fairly expensive. They are also based in the USA, which is a big downside for me.

The last option is Hetzner, although there are 2 options here: Storage Box or a dedicated server. I decided to list the dedicated server as an option here because dedicated servers have their own storage of course, meaning ZFS is a possibility, and the current price on the server auction with 5TB of storage is between 60€/month and 65€/month, which makes this option roughly equivalent to the cost of rsync.net.
Hetzner Storage Box is what I’m currently using, although with 1TB instead of 5TB, and it has served me well so far. It’s very cheap storage compared to the other options with only 12,97€/month and is accessible over SFTP, among some other protocols. Hetzner being headquartered in Germany is a big plus for me as well.

Deciding on the service?

Looking at all these options really made me realize that bulk off-site storage can get very expensive very quickly, and without thinking about it any more than this the obvious option is Hetzner Storage Box because it’s just so cheap, and keep on backing up my stuff with restic.

But these prices in general really do make me think about whether off-site backups of my “unimportant” media is necessary or not. What I’m sure about is that I don’t want to lose all my media files, but I’m not sure if paying up to $60/month is the solution here. I should instead ask myself whether off-site backups are what I want in the first place.

And so I’ve reached my conclusion for now: the answer is a simple “no”. I have enough spare hard drives laying around that I’m not using to back up a few terabytes, and I think this is what I’ll ultimately do: manual, local backups of the stuff that’s not so important it deserves an expensive off-site backup. Sure, I might lose a few songs or movies because I won’t manually back up everything very often, but none of this media is incredibly important to me, which is why it’s on its own dataset in the first place.
Because I already own the storage, I can instead do the manual backups and store one copy at a friend’s place, then swapping the old backup drive out with a new one every few months or so. This way I still have the upside of off-site backups without the downside of throwing money out of the window.

Not everything has to be perfect, and this is an example of a compromise that I can live with. Fully automated off-site backups are great, but they can also be very expensive for someone with a hobby.

Do you have a comment on one of my posts? Feel free to send me an E-Mail: witcher@wiredspace.de
To participate in a public discussion, use my public inbox: ~witcher/public-inbox@lists.sr.ht (Archive)
Please review the mail etiquette.

Posted on: May 24, 2026

Articles from blogs I read

Game of Trees 0.126 released

Version 0.126 of Game of Trees has been released (and the port updated). Complete release notes are as follows: Read more…

via OpenBSD Journal May 23, 2026

New blog design

I redesigned my blog! I decided to put some more personality into it this time, after over a decade of the minimalist style. This short post is just an excuse to show up in your feed reader so you can go look at it. Cheers!Also: I’m trying out…

via Drew DeVault's blog May 20, 2026

Project goals update — April 2026 (end of 2025H2)

The 2025H2 Project Goal period has now concluded. Over these months, the Rust Project pursued 41 Project Goals, 13 of which were designated as Flagship Goals. This post contains curated updates on our progress since the last post and the final status for …

via Rust Blog May 18, 2026

Generated by openring