Data Guard offers two main services:
1 Redo Transport Service: The Primay end of the redo log transmission to one or more standby destinations.
2 Redo Application Service: The redo log which is transmitted from the Primay end is applied to the standby end.
This article first discusses the redo transport services.
1, the use of ARCN transmission redo log
By default, the ARCN transport redo log, but only in the most high-performance mode can use ARCN (specific reference to the [Oracle] Data Guard Three protection mode introduction ), using the arch transmission redo log diagram is as follows:
The approximate process is as follows:
1 Primay segment ARC0 Once the log switch is completed, ARC1 transmits the newly generated archive log to the standby end;
2 standby driven by RFS process to accept the log, if configured standby redo log, log to standby redo log, and so standby redo log to do log switch to form an archive log, and then apply the archive log for recovery If you do not configure the standby redo Log,rfs process to receive the log, put it in the standby-side archive directory and standby reapply the archive log for recovery.
2, the use of LGWR transmission redo log
The use of LGWR process and ARCN is very different, the most obvious difference is that it does not need to wait for the primary completion of the log switch after the transfer, the diagram is as follows:
The process is generally as follows:
1 Once primary has redo log generation, LGWR will trigger LNSN process transfer redo only standby redo log; Note: This cannot be transmitted directly from LGWR because the entire database instance has only one lgwr, in order to ensure its primary performance is not affected, cannot be transmitted directly by it)
2 Network transmission mode can choose Sync or async,sync refers to when primary commits, must have to wait for redo transmission to standby successful, can return. So if you set sync, we recommend that you set the Net_timeout parameter at the same time, the timeout is not responding, then return an error. Async refers to the success of primary commits and the success of the log transmission, which has the least impact on the performance of primary.
3 Standby end of the RFS process to redo write standby redo log, if the real-time application, will be redo to apply to the standby database, if not open real-time applications, such as standby redo Log is archived and then applied to the standby database.