That's like comparing rsync to google drive. One is an open source tool where you can use whatever back-end you want, the other is a service. (Which is fine, just different kinds of things.)
However, in this case it's the open source tool that has a much easier user interface (I am actually proficient with tar, but still my tarsnap experience is like comparing 'restic backup /my/files --repo /mnt/backupdisk' with https://xkcd.com/1168/)
Indeed. Restic is just something you apt install and nobody provides you any service (you have to organise your storage space yourself); tarsnap is not simply free to use for yourself with your own storage. (Not saying it has to be free, but that's what makes it the definition of a service you have to purchase.)
Tarsnap works out as ~6$/GB/year for me. That’s for a mostly managed backup service. The only thing missing is snapshot pruning which is slow and a bit of pain due to the way tarsnap’s cache works. Restore is on par with restoring from tape — reliable but slow, but who can really complain about how fast disaster recovery is?
Raw managed storage with rsync.net is 0.18$/GB/year.
Do it all yourself, with the associated peril and time sink that entails, and the disks will cost you 0.04$/GB/replica.
Tarsnap has its place and I’m still a happy customer, but it’s one small part of a wider strategy that includes bulk storage elsewhere — rsync.net with borgbackup and plain rsync, on premises ZFS dumpsters, and offsite drives used like they are tapes.
Is the Microsoft OneDrive is $70 for 1TB a good option for the truly paranoid, or Microsoft will share the content in case they get a court order or some with a tank in their front door?
Tarsnap is not expensive at all for its target audience: folks with highly compressible data. For any data that's not very compressible, it's super costly.
Example, if you are into photography it's not uncommon to generate hundreds of GBs of files _per year_. Only in 2020, I generated over 200GB of photographs. Putting that on Tarsnap would cost me about $60/month. In 5 years time, I could be paying upto $4000/yr. Tiered services like B2 would cost an order of magnitude less.
Why on earth someone would pay 10x-20x alternatives for encryption that these days is available in high quality free open source software such as Restic, Borg or Duplicacy?
Previously on HN:
https://news.ycombinator.com/item?id=21410833