Hi Jeff,

Thanks for your help, but I believe I can't use your "modus operandi" in
my case. 
Reason is following : if our server crash happens during the night,
european time, the hacmp function will automatically transfer db content
on another node, and users will continue working, because it's daytime
in the us or in asia, thus generating lots of loglogs. When our team
will be back, let's say 12 hours later, and try to restore this db on
it's original node, all the work done during the night will be lost if
we can't restore it (like you said "data accidentally backed up under
the wrong node name is most likely not retrievable by onbar though or
very difficult.") That's where it hurts !
Do you understand what I mean ? (sorry my english may not be perfect
...)
So, what I'm looking for, is a backup strategy that works without losing
any snippet of already typed-in info, I can't afford losing 12 hours
work all around the world...
Add to this situation is even worse than described, due to the fact our
hacmp "backup" node is hosting an other informix db .... but that's an
other story !
If you have other proposal, my eyes are wide opened !
Thanks anyway, your info helped me in some ways ;-)
Cheers.
Arnaud