[sdotsen]

Why not S3? I use Arq with my Mac to back up my important doc, which gets encrypted in my S3 bucket. However, I'm still looking for a viable method to backup all my photos and videos. Right now they reside on multiple external hard drives.

[rsync]

"Why not S3?"

S3 is a great solution and we have nothing negative to say about it. I will note that our special 'borg' pricing gets you down to 3 cents, and we charge nothing for usage/bandwidth/transfer, so pricing-wise, we're about the same.

The main thing that differs (and this is important to some people) between S3 and rsync.net is that rsync.net gives you an honest to god blank unix filesystem to do what you please with.

So while there are many very nice tools that target S3 natively, you can't do magical things like this:

  pg_dump -U postgres db | ssh [e-mail] "dd of=db_dump"

  ssh [e-mail] "git clone git://github.com/freebsd/freebsd.git freebsd"

  ssh [e-mail] "test -f fileThatExists"

  ssh [e-mail] du -Ahd2 some/directory

Although it's worth noting that we do maintain 's3cmd' in our environment, so you can do things like this:

  ssh [e-mail] s3cmd get s3://rsync/mscdex.exe

[StavrosK]

You need this:

https://www.stavros.io/posts/holy-grail-backups/

About your S3 question, having something like rsync.net (where you can run a server process) is much faster/easier/better because your backup software can know exactly what's stored on the remote side and only send diffs.

I'm sure this can be overcome with some clever client-side engineering, but having a server process is always going to be more flexible overall.

A more pragmatic reason is "because Borg doesn't run as well on S3".

[cvwright]

This is very helpful. Thank you!

[StavrosK]

No problem, glad I could help!

[ooqr]

http://arstechnica.com/information-technology/2015/12/rsync-...

[leni536]

I never used git-annex with an S3 remote, but generally I have good experience with git-annex.