You Asked
My Prod environments is like this.
Three Node RAC, Active Data guard enabled.
There is a partitioned table, month based range partition, every month data will be in one partition. Each partition size is 180G to 200G.
I have another table for archival purpose. We will move 3 partition to archival table from main table every quarter.
After moving a data to archival table, the index rebuild will take longer time.if drop and recreate index will also be like that and take very longer time.This archival table is also used by application.so index is must. It is global index.
To move a data from main to archival table and index rebuild on archival, i am afraid to go for NOLOGGING option. If would use this,whay will happen to my SECONDARY db. Will it be sync because no logging and no redo.
My window to do this activity is 6hrs( i can avoid application connection).
What is the best method, kindly advise.
Three Node RAC, Active Data guard enabled.
There is a partitioned table, month based range partition, every month data will be in one partition. Each partition size is 180G to 200G.
I have another table for archival purpose. We will move 3 partition to archival table from main table every quarter.
After moving a data to archival table, the index rebuild will take longer time.if drop and recreate index will also be like that and take very longer time.This archival table is also used by application.so index is must. It is global index.
To move a data from main to archival table and index rebuild on archival, i am afraid to go for NOLOGGING option. If would use this,whay will happen to my SECONDARY db. Will it be sync because no logging and no redo.
My window to do this activity is 6hrs( i can avoid application connection).
What is the best method, kindly advise.
and we said...
Typically databases with a DataGuard setup have the 'force logging' mode set, so nologging requests are done with logging anyway. So that will be the first thing you would to check.
You *can* do nologging operations, but once completed, the tablespace/datafiles affected will need to be re-copied to the data guard node(s) because thats the only way the changed data can be seen (because it is not in the redo logs).
See
You *can* do nologging operations, but once completed, the tablespace/datafiles affected will need to be re-copied to the data guard node(s) because thats the only way the changed data can be seen (because it is not in the redo logs).
See
http://docs.oracle.com/database/121/SBYDB/scenarios.htm#SBYDB00920for the steps on what to do. Does that 6 hrs apply to your active DG database as well ? Because if it does, it might take longer than that to re-copy those datafiles.