Good questions. I have also given a lot of thought to those scenarios,
particularly
with regard to the fact that we're moving away from the federated model (since
no one ever used it, as far as I can tell). Citadel is finding itself most
comfortable in the space somewhere *between* the "just use gmail or o365"
people and the "deploy 20 Exchange servers and a big staff to maintain them"
people.
You're probably already aware that the latest Citadel server has the system
configuration, state tables, user profiles, and user photos moved into the
database. My goal is to move 100% of everything into the database. It will
come as no surprise that once we get there, we can do things like high
availability
using the replication features that are built in to Berkeley DB, or log shipping
etc.
But obviously we're not *quite* there yet. If you want to "fake it" and
still have a supportable
system, I would say go with one of the installation methods that puts everything
in /usr/local, and then mount /usr/local over a DRDB volume. Not only will
that be transparent to Citadel, but it ought to be able to transition cleanly
to whatever we build next.
