This figure illustrates the main actions of application continuity when it restarts and recovers an application. In the figure, a client application makes a request that is passed to a middle tier connection pool, such as the Universal Connection Pool (UCP),WebLogic Server, or JDBC thin driver. The requests from the client are sent to the database as calls. A planned or unplanned event or recoverable error occurs, which results in a potential interruption. Fast Application Notification or Fast Connection Failover terminates the session and Application Continuity begins to replay the session.
Application Continuity replaces the dead session with a new session and prepares to replay the transaction by using Transaction Guard to determine the outcome of the in-flight transaction, if one was open. The FAILOVER_RESTORE configuration parameter restores the initial session state and Application Continuity rebuilds the database session, recovering the transactional and non-transactional states before ending the replay and returning to run-time mode, submitting the last queued call. The response is then sent back to the client, which can continue if replay was successful, or must handle the error if it was an unrecoverable error.