[rdiff-backup-users] Deleting incrementals and saving resources

[Greg Nolan]

Hey all,

I have a question about deleting incremental backups. Here is my scenario:

We purge incremental backups older than 30 days. Say I run my first full on the 
first of march; in that full is a file called file.txt. Before the backup on 
the 2nd of march, file.txt is changed. On the 2nd of April we run a command 
which purges the incremental from the 2nd of march (deleting the changes in the 
backup system). That night we run an incremental which grabs the changes of 
that file (the differences b/t the first full and current backup).

I am hoping my understanding of the --remove-older-than option is accurate.

My questions:

1. does this mean that before the backup on April 2nd and after the purge on 
April 2nd, the changes made to file.txt are not accounted for in my backup 
solution? 
2. is there any way to merge the full and incremental backups (on the server) 
to a point in time to prevent this?

This brings up a problem I'd like to avoid. Say a backup client has 6gb of 
static data on the full and creates 20gb of static data in a month, each month 
I have to send AT LEAST 20gb of data over the wire (which gets saved then 
deleted when it expires). If the data could be merged on the server (rather 
than deleted), it would save unnecessary use(read/writes) of disk and bandwidth.

Input, suggestions, and ideas? 

thanks,

-g