[XCSSA] From SANS: "So.. Are all of the bad guys really on the outside?"

[X-otic Computer Systems of San Antonio]

On Feb 6, 2008 12:23 AM, X-otic Computer Systems of San Antonio
<xcssa at xcssa.org> wrote:
>
> Sounds cool!  I've wanted to start looking at using S3 based diff backups.
>
> Tweeks
>

I'm using duplicity (http://duplicity.nongnu.org/) to backup my web
servers to S3.  Duplicity is an incremental tool, using rdiff, that
stores the backups in 5MB tar.gz files on the remote end.  You'll need
the S3 library for python, and the latest duplicity build or check it
out of svn to get S3 support.

I backup each of the home dirs to a bucket, plus mysqldumps of the
databases.  I have successfully restored a number of files, trees, and
even a complete restore a few times.  One concern is that as the
increments grow, so do the duplicity index files and S3 transfer
charges.  Duplicity has an option to force a full backup and delete
old sets which I need to explore using once every 3-6 months.

My usage for January was 35 GB-Mo storage, 4GB in, and 36 GB out for a
total cost of $12.33.  A good chunk of the 36GB out (about 20GB)  was
a full restore of one of my web servers and the rest was duplicity
reading the index files.

If anyone is interested in the shell script that I use for nightly
backups let me know and I'll clean it up and comment it.

--anton