Protection mode decision is crucial and database administrators need to work with IT managers and other responsible people to determine RTO (Recovery Time Objective) and RPO (Recovery Point Objective) values and to select the most appropriate mode for their Data Guard configurations. After the decision is made, setting the data protection mode is a simple operation that can be performed by SQL*Plus, Data Guard command-line interface (DGMGRL) or Enterprise Manager Cloud Control.
Data Guard offers three data protection modes, which meet different business requirements as mentioned in Chapter 1, Getting Started.
The following are the different modes:
- Maximum Protection
- Maximum Performance
- Maximum Availability
Let's look at the details of these protection modes and see how we can switch between the different modes.
The Maximum Protection mode is referred to as the Guaranteed Zero Data Loss configuration. A primary database operating on the Maximum Protection mode doesn't provide an acknowledgment to the users that the commit is completed until transactions are successfully transferred to at least one standby destination. This setup requires the SYNC redo transport service using the LGWR attribute and guarantees that no data will be lost on the standby database in case of a primary database failure.
Of course, guaranteeing zero data loss comes at a cost. Because the primary database will always wait for an acknowledgment from standby destinations to continue its operation, there will be performance implications on the primary database. However, with 11g, the performance effect of using the SYNC redo transport service is less than the earlier releases. In the previous releases, the primary database doesn't send a redo to the standby database before completing the write to online redo logs. In 11g, the database writes redo to online redo logs and sends it to standby destinations simultaneously. This behavior reduces the time waited to complete a commit for a primary database.
Consider the following points before setting the Maximum Protection mode:
- Network bandwidth between sites is essential in this mode. If the bandwidth and latency of a network fails to satisfy real-time transport of redo generated by the primary database, there will be serious performance- and database-availability problems on the primary database.
- Using more than one standby database (preferably a physical standby one) for a Maximum Protection configuration is a good practice, which will increase the uptime of the primary database on standby and network failures. Also, the data protection will continue even if you lose the primary database and failover to one of the standby databases. It would be better to locate each standby database on different locations if possible.
- The primary database must be on the mount mode when changing the data protection mode from Maximum Performance to Maximum Protection.
- On all standby databases of the Data Guard configuration, the standby redo logs need to be created with the correct number and size before using the Maximum Performance mode. It's also a good practice to create standby redo logs in the primary database in order to be ready for a switchover.
- It wasn't possible to use a logical standby database with the Maximum Protection mode before 10g, because standby redo logs weren't supported by logical standby databases. Starting with 10g, we're able to use a logical standby with the Maximum Protection mode; however, we have to consider unsupported data types in such a case.