The failover strategies that the RCM supports are: manual, automatic (switch active), and failover with Replication Server quiesce.
Manual – the RCM notifies the system administrator of a failure, but does not control processing at the Replication Server. The administrator must control the Replication Server.
Automatic – the RCM coordinates failover automatically and the direction of replication is reversed. This means that the roles of the standby and the active databases have switched. The standby database is now the active database, and changes are replicated from it to what was originally the active database and is now the standby database.
Failover with Replication Server quiesce – the RCM coordinates failover automatically. The Replication Server is quiesced and replication is stopped. When the Replication Server is quiesced, it does not capture changes in the standby database.
Choose manual failover to control every step of a failure and manually rebuild the active Adaptive Server rather than have transactions automatically applied to the standby database after failover.
Choosing between automatic failover and failover with Replication Server quiesce depends on the volume of data coming through the system, the size of the Replication Server queues, and the length of time you expect the active server to be unavailable.
If you choose automatic failover and switch the warm standby connection, the Replication Server captures all transactions entered at the standby Adaptive Server and stores them so that they can be applied to the active Adaptive Server when the server recovers. This simplifies the recovery process because the active databases do not have to be reloaded. However, if there is a heavy volume of data or the active Adaptive Server is down for an extended period of time, the Replication Server’s queue might not be capable of storing all of the transactions. If this is the case, choose failover with Replication Server quiesce.
