Why I use Arq instead of restic for macOS backups
I recently finished upgrading all my Macs to the long-awaited version 7 release of my preferred backup software, Arq. I use it to backup all my Macs and Windows computers to the cloud storage provider Wasabi, which is inexpensive relative to eg. AWS S3. I use the open-source backup software restic to backup my Linux machines, also to Wasabi.
(Yes, I know AWS and GCP now have cheap archival storage options; but I’m not migrating terabytes of backup data.)
During a discussion about Arq 7, Andrew asked why I don’t just use restic for my Mac backups as well. This is a lightly-edited version of my answer:
I actually have a few reasons for this:
To use restic, I’d have to wrap it in my own backup script and run it via launchd myself.
I would have to to extra work to get eg. notification emails on errors, which Arq handles for me out of the box. And I know from experience I’d spend a nontrivial amount of time working through macOS permissions problems and other launchd weirdness. Plus, I don’t even know how I’d run restic on Windows, though that’s a minor concern.
Arq has controls for limiting network bandwidth, CPU usage, and (new in Arq 7) disk IO; and it sounds like Arq 7 has added some performance improvements that I wouldn’t get with restic.
CPU and bandwidth limits can be achieved, of course, with standard Unix tools, just like email notifications, but again: I don’t want to rewrite Arq.
The file restore experience is easier with the Arq GUI, especially with its drag and drop integration with Finder. I’ve restored files with both restic and Arq; it’s possible with restic but easier with Arq.
I don’t trust restic quite as much.
I have run into weird errors which imply that my restic repository has had some nonzero number of integrity issues. The restic devs were able to help me prune out bad backups and get the repo back into a healthy state, and I think they fixed a possible cause for this issue in the most recent version, but … still.
That all said, I’ve been able to restore files on multiple occasions from that and other restic repositories.
I wouldn’t gain a lot.
I’d do a lot of work getting all the aforementioned features & behaviors working smoothly, and at the end of the day I’d have a DIY version of Arq that I trust less than Arq. I’d be backing up to the same backend, so cost would be similar. Only difference is the cost of Arq, which I’ll happily pay not to have to do all this myself.
At the end of the day, if the Arq team released a Linux version of Arq that somehow worked as well on the server as it does on macOS, I’d consider moving to it from restic.