DATA GUARD主库和备库之间为什么只能进行事务级别上的一致，有什么说法吗？

Yong Huang

原帖由 jerryma6002 于 2011-2-16 20:02 发表
in that case, redo info will be still transfered to standby site.  but applying process ( MRP ) at standby site will ignore this failed transaction.  

this should be not new feature because  it  conform to oracle read consistency mechanism ( such as read commited data ) .

I have some doubt about this. If you can provide some document saying the same, that would be great.

Yong Huang

jerryma6002

sorry, so far, no public official doc is found to describe this directly. that is deduction based on read consistency feature.  

From designer's viewpoint (at algorithm level ), MRP should work like this due to consideration of efficiency. do u think so ?

[ 本帖最后由 jerryma6002 于 2011-2-18 10:02 编辑 ]