(For home-grown options, I've been using rsnapshot for years. But it really works best on a local disk and the rsync-over-ssh hacks I've done are the opposite of "clean consumer solution".)
The killer feature Duplicity has for consumer use is that it's probably the only (?) backup program on Linux that has an actual quality desktop GUI called Déja Dup [3] that's included by default in Ubuntu. It works somewhat similar to how Time Machine works on Macs. But if you need to backup headless systems the Duplicity command line interface works fine as well.
When it comes to backup features it's not the most powerful tool compared to some other solutions like Attic or Borg, but I think the GUI and out-of-the-box (encrypted) integration with cloud services make it one of the most user friendly solutions on Linux.
[1]: http://duplicity.nongnu.org/
[2]: https://www.backblaze.com/b2/cloud-storage.html
[3]: https://www.linux.com/learn/total-system-backup-and-recall-d...
[3b:] EDIT: Also Duplicati has a nice web GUI. Worth looking into as well.
Modern incremental-forever backup tools like borg[1] (fork of Attic[2]) are much faster since they are based on block hashing, which also gets you deduplication for free.
Basically, a backup is a set of hashes - this means that you can selectively delete or retain old backups without having to merge them.
I built a >50 TB enterprise backup cluster with borg and it works extremely well.
E.g. if one backs up to hubiC, OneDrive and Google Drive, with a RAID5-like redundancy, they can have just 33% overhead and still be sure shall any of those vendors discontinue the service or suffer a failure, their data would be still safe. Some call that RAIC - Redundant Array of Inexpensive Clouds.
git-annex and Tahoe-LAFS do this, but they're not actual backup solutions.
If you go with this one, do be aware that by its nature, restores when you've done a lot of incremental backups can be extremely slow. (Think hours, where other systems would retrieve the data in minutes if not seconds.)
That means you really need to be a doing a full, and therefore also slow, backup reasonably frequently to keep Duplicity usable. As a rule of thumb, we found a full backup each month and dailies each night was just about tolerable even if you needed a restore quite late in the month, but even then a 10GB backup with 30 days of incrementals on top could be looking at an hour or more to retrieve a single file.
As I understand it, Duplicity's scheme can also be vulnerable to minor corruption (e.g., due to disk error) anywhere since your last full backup taking out the entire backup, so if you're going to use it then you might like to investigate the options it provides for mitigating this risk as well.
If my understanding is correct Duplicati started out as a C# rewrite of Duplicity (hence the name), but the projects have grown apart and are different things now. I don't think you can use Déja Dup with Duplicati, for example.
Is it a proper service on Windows? That's a problem. I want to configure it from an admin account and have it still work when non admin/non tech savvy users login.
There's also restic, which works similarly, but has native support for cloud storage: https://restic.github.io/ - sadly it's claims of being efficient are rather undermined by the continued lack of compression support.
rsync.net has some kind of problem with the "Powerboost" feature. The first 10 megabytes of a transfer are fast as hell, then it drops to ~150kB/s. Have fun backing up gigabytes at that speed. There is something wrong with their traffic shaping. After much back-and-forth their tech support basically threw up their hands and gave up.
Dropbox takes 9.9 USD per month for 1 TB
I have a mirrored 2x6TB setup that all of my computers back up to. It shows up as an Apple TimeMachine, so MacBook backup and restore is easy; and then it also some media storage with automatic nightly/weekly/monthly snapshots - the monthly ones hang around for a year, the others are cleaned up sooner. For those, I have to un-hide the snapshot folders, but then I can just copy-paste any deleted files that I want to restore.
It definitely takes some setup work, but I've been pretty happy overall.
That server is in my basement, my work laptop also has CrashPlan for off-site backups, but that's a business plan, so it won't be affected by this. I do need to figure out an off-site backup for the rest of it...
rsync.net
It's not the cheapest either, but I never had issues when I was using them.
And, frankly, my measly 200GB isn't much data. A 2TB HDD costs under $100 and could cost $500/mo to back up with tarsnap. It would still have been $5/mo with CrashPlan.
I get that Colin is basically just charging the S3 costs plus a small margin, but the reality is that S3 is far too expensive to be used to store low-value consumer data.
The majority of my most important backup data is my photos. I don't see how tarsnap could duplicate (or even further compress) that.
I have approx 2.5tb on CrashPlan, so that gives me an upper bound of $625 per month. Maybe it might be 50% lower when deduped, maybe more, maybe less. Who knows. Not really a pricing model I can make a decision on.
Tarsnap looks great for certain use cases. But it doesn't really make sense for home users with a large amount of data.
Using the --dry-run flag.