A few words about rsync.net
A month or so ago, I decided that it was long past time to come
up with a more robust backup solution than the ad-hoc methods
I was currently using. Mainly this involved some dropbox-like
storage services such as Google Drive and Box.com. For larger
things, this "strategy" involved random usb keys and microSD cards.
However, it was hard to find anything I really liked. Some solutions
were dependent on a client running on a particular machine, which limits
flexibility on platforms. Others that are typically suggested for this purpose
(Glacier, etc) have very high costs (both in money and in time) for
actually retrieving your data. Still others (S3 and clones), while they
didn't require a client and didn't charge more (or much) for transfer,
still made you write tools for a particular (non standard) API. You
might argue that S3's API is so widely adopted among storage systems now
that it's a de-facto standard, but that's not really good enough for something
I want to last a lifetime.
What I really wanted was ssh access to a machine where snapshots are automatically
performed on disk and everything is managed for me. This is what rsync.net [1] is.
You get a FreeBSD chroot/jail ZFS filesystem that you can access using standard tools
over SSH. You don't get a shell, but you don't need one - just add commands
after SSH: `ssh foo@bar mkdir test`.
The pricing is up-front and reasonable and includes daily snapshots.
There are no extra costs for getting your data and crucially, you
talk directly to engineers when emailing support. Git is installed
on the remote machine so you can also use it for private Git repos
if you desire.
I've been very happy with it when paired with git-annex, which will probably
be the subject of a blog post on https://jbo.io
That's all for now!
[1] https://rsync.net
