Proposed description of Backshift

[Saint Germain]

Hello,

As I have explained, I am writing an article on open source backup
software which include deduplication.

Here is what I proposed to decribe Backshift.

Backshift is visibly developed by someone who love Python (see the
different interpreter tested and the performance tests). The emphasis
is put on archive size optimization (file data chunk deduplication and
use of LZMA for compression).

Here are a few remarkable points :
- File data chunk deduplication on the client
- Parallel/concurrent use possible by several users (!)
- Resume backup after interruption possible
- No data encryption
- Possible use of sshfs to secure the connection
- Initial creation of a lot of directories/files but which allow in
theory to optimize the total size on a lot of backups.

I would be very interested to have your comments.
Sorry for the quick translation from french !

Regards,

[Dan Stromberg]

Sorry it took me a while to get back to you.

Your description is most satisfying.  Thanks.

I might add that it doesn't require a dedicated server process it operates over a typical network filesystem like NFS or CIFS or sshfs, provided that the underlying server filesystem can store 2^20 subdirectories in a single directory.  ext4 and xfs do this well, ext3 and netapp do not.

--
Dan Stromberg