In my current setup, we have 8 servers that update a single server as the data is entered (which server the user logs to is transparent to them and is controlled by entries in an address table) The advantage is that the user only send the data to one location and that location deals with problems in transferring the data (single largest problem, replication breaks and we must diagnose, fix and re-start replication). You need to be aware how SQL server does updates (and deal with it accordingly), internally, as that's currently the main breaker of replication in our scheme.