1.)

The backup details box shows:

disk (ext3) - 1329MB/12032MB

df -m on the linode:

/dev/xvda 11657 1101 10556 10% /

What accounts for the difference in size?

2.)

If I need to restore would I restore to free space in a linode's image or restore to a partition that I had already created? I guess I'm asking is this a full image snapshot or is it basically a tar of all the files (without partition info)?

3.)

Are there any plans to allow downloading of these snapshots for ultra safe backup? I realize things like fire in datacenters are rare events, but I'd still like to keep a fairly up to date local copy.

Re your #2, I got a chance to play with it on the development server last month and I had to temporarily "purchase" additional space to restore the image in to so I could mount it and copy off what I needed. In the case of a catastrophic restore (like you 'rm -rf /' as root) you'd probably just delete your original image file and just restore it.

The restore produced a new disk image that was all ready to boot from.

As for #3, I intend to still rsync a backup off-site anyway, as I do now. That will still be easier for individual file restores anyway. If you have the ability to copy an entire image off-site somewhere, then doing so by-file shouldn't be that big of a deal I'd think.

It's good to have multiple methods of backup.

2. You do need disk space on the restore target Linode, but the restored disk images are only as big as needed for their contents and not as big as the original disk images.

The restore recreates the profile and any disk images (and links them together), all prefixed with "Restore #" so they're distinct from any existing images. So you have a ready-to-run profile; if you have enough space the original profile and its image(s) can still be on the linode, intact.

3. Being able to download a compressed snapshot would be nice.