We currently have a production database 11.2.0.4 (A) and an active physical standby (PSA). we are in the process of developing migration code for this database because it is vendor supplied schemas. The target database is 12.1.0.2 (B). when we go to do this in prod, this is what we were planning to do
1. Stop all traffic to database A and point read only traffic to PSA.
2. Run the migration from A to B
3. After the migration is complete, stop traffic to PSA and reroute to database B
4. Create DR for database B
In the above scenario when we are in 1, we are operating without a DR. if the server for database PSA goes down we have full outage. In this scenario is there something else better that we can do ?
There is no limit to the number of standby databases you can have, with the caveat that you need to have a licensed server. So you *could* do something like:
0. Create PSA#2, leave PSA#1 in place
1. Stop all traffic to database A and point read only traffic to PSA#2.
2. Run the migration from A to B, archivelogs etc still get sent to PSA#1
3. After the migration is complete, drop PSA#2 and reroute to database B
4. DR for database B is already done, because PSA#1 has been updated during the exercise
This of course depends on what you mean by "migration". I'm assuming a standard upgrade from 11 to 12 here.
If it was something more significant, then you could tackle things via a different route, that of minimising the outage window to almost zero using transportable database.
An *outline* of the process is here on my blog
https://connormcdonald.wordpress.com/2015/06/06/cross-platform-database-migration/ but there are more detailed instructions on MOS.