Selfhosted

My friend has 1G/1G Internet. I have a rsync cron job backing up there 2 times a week.

It has a 8TB NVMe drive that I use bulk data backup and a 2TB os drive for VM stuff.

Most of my work is with Macs, and even one server is running macOS, so for those who don't know how it works 'over there', one runs Time Machine which is a versioning system keeping hourlies for a day, dailies for a week, then just weeklies after that. It accommodates using multiple disks, so I have a networked drive that services all the mac computers, and each computer also has a USB drive it connects to. Each drive usually services a couple of computers.

Backups happen automatically without interruption or drama.

I just rotate the USB drives out of the building into a storage unit once a month or so and bring the offsite drives back in to circulation. The timemachine system nags you for missing backup drives if it's been too long, which is great.

It’s not perfect but very reliable and I wish everyone had access to a similar system, it's very easy, apple got this one thing right.

LTO8 in box elsewhere

The price per terabyte became viable when a drive was on sale for half off at a local retailer.

Works well and it was a fun learning experience.

I built a near identical server for my parents and just sync my nextcloud folder to theirs using syncthing

I bring 1 of my backup disks to my inlaws. I go there regularly so it's a matter of swapping them when I'm there.

I also had been contenplating this for a while. The solution I implemented recently is:

The system itself is a RPI on NixOS. The system can be reproduced from the NixOS configuration. The NixOS configuration is stored on GitHub. Since I can reproduce the sdcard image (and full system) from the configuration I opted to not do any backup of the sdcard/system itself.

I've also opted to not use raid, as I can replace/add a RPI without too much hassle.

The real backups for me are for photos. Those are stored on a M.2 storage. A second (similar) RPI is placed at my dad's place. The rpis run tailscale and syncthing. Syncthing syncs using staggered mode (stores 1 version for the last day/week/year) and the RPI at my dad is untrusted, so the backup files are sent/stored encrypted there.

This setup hasn't run very long yet, so I won't recommend it, but it seems to check quite a lot of boxes for me. Maybe it gives some ideas. I'm also interested what alternative solutions others came up with.

I tend to just store all my backups off-site in multiple geographically distant locations, seems to work well

My automated workflow is to package up backup sources into tars (uncompressed), and encrypt with gpg, then ship the tar.gpg off to backblaze b2 and S3 with rclone. I don't trust cloud providers so I use two just in case. I've not really been in the need for full system backups going off site, rather just the things I'd be severely hurting for if my home exploded.

But to your main questions, I like gpg because you have good options for encrypting things safely within bash/ash/sh scripting, and the encryption itself is considered strong.

And, I really like rclone because it covers the main cloud providers and wrangles everything down to an rsync-like experience which also pretty tidy for shell scripting.

I have a storage VPS and use Borg backup with Borgmatic. In my case, I have multiple systems in different repos on the remote. There are several providers, such as hetzner, borgbase, and rsync.net that offer borg storage, in the event you don't want to manage the server yourself.

Veeam Backup&Replication with a NFR license for me.
My personal setup:
First backup: Just a back up to a virtual drive stored on my NAS
Offsite backup: Essentially an export of what is available and then creates a full or incremental backup to an external USB drive.
I have two of those. One I keep at home in case my NAS explodes. The second is at my work place.
The off-site only contains my most important pieces of data.
As for frequency: As often as I remember to make one as it requires manual interaction.

Our clients have (depending on their size) the following setups:
2 or more endpoints (excluding exceptions):
Veeam BR Server
First backup to NAS
Second backup (copy of the first) to USB drives (min. of 3. 1 connected, 2 somewhere stored in the business, 3 at home/off-site. Daily rotation)
Optionally a S3 compatible cloud backup.

Bigger customers maybe have mirroring but we have those cases very rarely.

Edit: The backups can be encrypted at all steps (first backup or backup copys)
Edit 2: Veeam B/R is not (F)OSS but very reasonable for the free community edition. Has support for Windows, mac and Linux (some distros, only x64/x86). The NFR license can be aquired relatively easy (from here and they didn't check me in any way.
I like the software as it's very powerful and versatile. Both geared towards Fortune>500 and small shops/deployments.
And the next version will see a full linux version both as a single install and a virtual appliance.
They also have a setup for hardened repositories.

Amazon AWS Glacier

Edit: I was downvoted for this, but it’s genuinely a more affordable alternative to Backblaze whose finances are questionable.