We set up an SDE database with 3 versions and 1 2- way distributed version. We later decided that we would like some of the distibuted data that is in a feature dataset to participate in a topology.
I am searching for the best practices way to add the topology to the parent database with a little pain as possible. I would prefer to not have to recreate my distributed dataset. I have some ideas, but I would welcome suggestions from an experienced user.
Probably not what you are looking for but I stay away from feature datasets in the SDE environment. They have a tendency to misbehave. My experience (going back to v. 8.0) with multiple versions has been less than stellar as well: when it comes to reconciling it about drove me mad trying to unravel conflicts.
That said, my experience with replication has mostly been outstanding: you might want to look at one way replication to a file geodatabase and let the feature datasets and topologies reside there.
Just my $00.02....