>Without at least one dedicated server, it's going to be very tough, since you need something to arbitrate transaction journalling and manage the replication process.


Yes I was thinking that some sort of transaction log that is maintained on all online units and some sort of update/reconciliation process when a unit comes back online.

Updates to the transaction log would have to be sent to all online workstations.

The case that is hard to adjust for is when the units with the most current copies of the transaction log are not online when the units that need the updates come online.