[Oracle] Introduction of three protection modes of Data Guard _oracle

The data guard provides the following three kinds of protection modes:

1 Maximum protection mode (Maximum Protection)
"Maximum protection" here refers to the maximum protection of the data is not lost, that is, at least one standby and primary maintain real-time synchronization, but this is costly, that is, when a transaction is committed, not only to write to the primary section of the online redo log, There's also a standby redo log written to at least one standby. A serious problem is that when a standby fails or a network failure causes the log to not be synchronized, the primary database is shutdown, which is not acceptable in the production system, so this pattern is rarely used.

2 Maximum Performance mode (Maximum performance)
The highest performance is just the opposite of maximum protection, and primary is almost unaffected when a transaction commits, as long as it is written to the primary online redo log. The redo is then synchronized to the standby, which makes it impossible to guarantee primary and standby real-time synchronization. This pattern in the actual application is more common, after all, primary is not affected is the most important, if there are too many standby behind primary, DBAs can manually intervene in the standby end apply redo.

3 Maximum availability mode (Maximum availability)
The highest availability mode is between the first two, under normal circumstances it is the same as the highest protection mode, but once the standby fails, switch to the highest performance mode immediately, primary will not shutdown. If the model is applied in practice, it is necessary to pay attention to the effect of real-time application on the performance of primary.

Whichever pattern you use, use Log_archive_dest_n to configure it, and the table above shows the relationship between them. As you can see, maximum protection is identical to the highest availability configuration (because maximum availability is normally the maximum protection): You must use the LGWR process to transfer redo, the network transport mode must be sync, the disk write option affirm, and standby redo log must be configured.

and the largest performance mode is more flexible, that can be used LGWR, can also use the arch process transmission redo, the network can be a sync can also be async, disk IO write can be affirm can also noaffirm,standby redo log also dispensable.