[rdiff-backup-users] atomic increment files?

[Marcel (Felix) Giannelia]

Is there any way of making rdiff-backup produce single files as 
increments (say, by zipping them together when it produces them), 
instead of thousands of itty bitty files? One file per increment would 
make the task of moving old ones onto archive DVD a lot easier, and 
would make a lot less hardship for the target machine's filesystem, too. 
It probably wouldn't slow down restores all that much, as accessing an 
archive file's directory structure is likely faster than doing the same 
in a part of the filesystem containing many thousands of files per 
directory.

Presently, I'm trying to do a du -s on our backup directories and it sat 
there for over an hour without having printed the size of the first one. 
According to top, du was using 50% of the total memory. I know that 
there are statistics files which I could add together, but in this case 
I want to use du to be sure because there's a chance that there might be 
stray files in our rdiff-backup-data directory. Also, creating so many 
files that commands like du cannot even function is, in my opinion, 
incorrect behaviour.

~Felix.