The failover process moves database activity from a source database to a target database in the event of a failure in the source database or on the primary machine. Determining when to initiate failover directly correlates with latency and acceptable target down time. Once either of these criteria has failed, a decision as to whether to fail over to the target database must be made.
Another scenario where failover is more appropriate is the loss of the source database. Loss of the source database can be due to something as simple as the database shutting down and needing to be restarted. A more severe example is the loss of hardware, making access to the database impossible. In this instance, you must determine whether the database can be recovered in an acceptable amount of time. The acceptable amount of time is determined by your business requirements for having the database and application available. If you can afford 15 minutes of database unavailability, then this is your measurement for failover.
Keep in mind that in the event of a failure, data can be lost. For example, if the source database shuts down between BI write and AI write and the currently busy extent is applied to the target, it will have the data written to AI but nothing else. The source and target databases are then no longer synchronized.
