The Maximum Availability mode is the data protection mode that has the ability to run as a Maximum Protection or Maximum Performance mode depending on the accessibility of standby databases. In a normal operation where the standby is up and able to receive redo data synchronously, the primary database acts like the Maximum Protection mode and waits for acknowledgment from the standby database to complete transactions. However, the key point of the Maximum Availability mode is the behavior of the primary database when it's not able to receive acknowledgment from any standby database. It waits for a predefined period of time and if the connection cannot be established, the primary database continues its operation as a Maximum Performance mode database. The number of seconds that the primary waits before marking a standby inaccessible is defined with the NET_TIMEOUT attribute of the LOG_ARCHIVE_DEST_n parameter. The default value of this parameter is 30 seconds. In a Data Guard configuration with the Maximum Availability mode, the primary database does not stall for more than NET_TIMEOUT seconds if it's not able to access any standby database.

When the primary database is not able to connect to the standby database for NET_TIMEOUT seconds, it stops sending connection requests to the standby database and continues completing transactions. Then, the primary retries connecting to the standby immediately after every online log switch. We can use the REOPEN attribute to set the time (in seconds) for which the primary attempts to reconnect to the standby. When the connection is established, the missing archived redo logs will be sent to the standby by the ARCH process simultaneously with the online redo transport.

With its logic, the Maximum Availability mode provides zero data loss in a normal operation, and the primary database's availability is not at risk when there is no accessible standby database. On the other hand, what we sacrifice by using this mode will be the guaranteed zero data loss feature of the Maximum Availability mode and the performance independency of the primary database in the Maximum Protection mode. This mode uses the SYNC redo transport with the LGWR attribute as in the Maximum Protection mode, which has an impact on the response time of the primary database. So once again, the network bandwidth and latency are very important in this protection mode.

We can state that in a Maximum Availability mode Data Guard configuration, data is at risk only when two failures occur consecutively on the standby and primary databases.