> > My best site ever was an airport, where they use to do a switchover on a regular basis. Every last Sunday of the month they switched between the Tower DC and the Terminal DC. 3-4 times a year they found out that some interface had a flaw. Imagine what would have happened if the first failover was performed after 2 or 3 years and would result in 6-12 flawing interfaces. The question is about frequency of this routines. Imagine what could have happened if someone changed you environment between your regular switchover intervals? It's somewhat better to think about this interval in terms of requirements: - test after every environment change. If you manage to catch all change events then you are done. - if you believe you missed some changes then you can do as frequently as need to feel comfortable, it's more about psychology then. How big is the environment to monitor? Provided you can reroute sqlnet connection at one point (change IP/DNS of standby for example) then it is the database server machine only as far as database is concerned. - training of personal. Training on training environment is safer. However if you are sure that your DG solution works then why should you be afraid to perform regular fail/switchover trainings? It's as simple as an execution of one script under normal conditions(no redo gap or lag.) In that case I can agree that regular fail/switchover is ok. The rest of weird cases (redo log gaps or lags) should be trained on training environment, IMHO. The ugly point is that what really matters in this training it is the ability to handle those weird cases, execution of one script under normal conditions is a trivial training. Let's agree that fail/switchover training on production should not be too frequent. It can be enough just to merge this training with fail/switchover tests due to environment changes. Brgds, Laimis. Fyrirvari/Disclaimer http://www.landsbanki.is/disclaimer -- //www.freelists.org/webpage/oracle-l