Re: [BackupPC-devel] Storing the backups on AWS's S3

[Guillaume F. <gf...@lo...>]

Guillaume Filion a =E9crit :
> I'm reading about Amazon Simple Storage Service
> (http://aws.amazon.com/s3) and a lot of people are talking about using
> it as an off site backup system. I'm wondering how hard it would be to
> configure BackupPC to save its files directly to it.

The main problem with using S3 with BackupPC is that there's no rsync
support, so you basically have to resend the whole file every time. That
means slow backups and more expensive bandwidth charges.

However, I finally was able to sign into Amazon EC2 (1) and I've been
playing with it last week-end. Since there's no bandwidth charges
between EC2 and S3, setting up a small server on EC2 with rsync or
rdiff-backup could mean "offsite backup heaven" for backuppc users.

Here's how I imagine such a system:
You have a local BackupPC server (bpc) and a EC2 server image (ec2).
ec2 uses S3InfiDisk (2) to store the data directly into S3.

1) bpc fires up ec2 using ec2-run-instances.
2) bpc knows when ec2 has finished booting using ec2-describe-instances,
so when ec2 is booted, bpc starts rsync or rdiff-backup.
3) When the rsync is done, bpc shuts downs ec2 using
ec2-terminate-instances.

Ref:
1: Elastic Computing Cloud, a virtual server at 10 cents per hour
2: http://www.openfount.com/blog/s3infidisk-for-ec2

Cheers,
GFK's
--=20
Guillaume Filion, ing. jr
PGP Key and more: http://guillaume.filion.org/