12.2新特性，ADG是RAC环境，在多个实例可以启动MRP进程recover，命令如ALTER DATABASE RECOVER MANAGED STANDBY DATABASE DISCONNECT FROM SESSION INSTANCES ALL
该特性只适用RAC和RAC one node环境数据库,不支持块跟踪,不支持In-Memeory列存储
注：MRP进程只在执行命令的节点运行，其他节点会产生pr* recovery processes进程。
Setting Up Multi-Instance Redo Apply
As of Oracle Database 12
Release 2 (188.8.131.52), a new
INSTANCES [ ALL |
clause is available on the SQL
ALTER DATABASE RECOVER MANAGED STANDBY DATABASE
It has the following restrictions:
The clause is applicable only for Oracle Real Application Clusters (Oracle RAC) or Oracle RAC One Node databases.
Block Change tracking is not supported.
In-Memory column store is not supported with multi-instance redo apply in an Active Data Guard (ADG) environment.
option causes redo apply to run on all instances in an Oracle RAC standby database that are in an open or mounted state at the time recovery is started. .All instances must be in the same state — either open or mounted. A mix of states is not allowed.
option restricts the number of instances that redo apply uses to the number you specify. For integer, specify an integer value from 1 to the number of instances in the standby database. The database chooses the instances on which to perform Redo Apply; you cannot specify particular instances.
view is only populated on the instance where recovery was started (where the MRP0 process resides).
If you omit the
clause, then recovery happens on only one instance where the command was issued.
Because recovery processes ship redo among instances, redo apply performance is directly related to network bandwidth and latency.
来自 “ ITPUB博客 ” ，链接：http://blog.itpub.net/10271187/viewspace-2637644/，如需转载，请注明出处，否则将追究法律责任。