[Duplicity-talk] Backup methodology

[Grant]

I've incorporated duplicity into my backup scheme like this:

I need to backup system A and system B.  Each of those systems
maintains a local duplicity repository of its important files along
with an md5sum of the archives in its repository.  Every night, system
A and B upload their repository archives and md5sum to system C and
download the other backed-up system's repository and md5sum, diff the
downloaded md5sum against a calculated md5sum, and email me the
result.

Is this a good methodology?  Is there room for improvement?

I've read about the verify function but I'm not sure if it would be
beneficial to incorporate it here.  If the contents of my files are
changing so that I can't use verify along with --compare-data, what
assurances does it offer?

Would it be beneficial to include a dummy file in my archives and
verify that file nightly against its known contents with
--compare-data?

- Grant