I'm curious about the stuff that's hard to replicate. Don't most of them store their stuff in files? Could those users perhaps use a distributed filesystem for replication? Clustered filesystems were used in supercomputers, are used by Google, and similar methods go back to VMS systems on VAXen.

I'm not saying something off-the-shelf will work or out of the box. Just seems like there would be a clear path for admins or developers of such software to integrate it with... something [1]... that did that.

[1] Example: http://sector.sourceforge.net/