0
votes

I would like to plan and test my database recovery in another site (another instance on another server in disaster recovery site). I take a monthly RMAN level 0 image copy every month and daily incremental level 1 backups. The database is running in noarchivelog mode. The online redo logs are multiplexed to a disk in the disaster recovery site. Also we have a recovery catalog on another server.

I want to test restoring the recent (yesterday) backup to database in disaster recovery site and then recover to just apply the online redo log files, how to achieve that?

side question: Is it sufficient to recover if we only have a yesterday backup and the online redo logs containing all transactions of today and none of them was overwritten? Since the database is in noarchivelog mode. What is the use of archivelog mode if we have a daily backup and the redo logs are not overwritten during the day until the backup is taken? what is the use of backing up archive logs?

1
Point in time recovery is only possible with archivelog mode. In noarchivelog mode all backups are cold backups; you can not recover anything from the online redo logs if you are performing an actual rman recovery and not just an instance startup automated recover (as following a db crash or restart).pmdba

1 Answers

0
votes

You are working with a dangerous setup since you seem to be betting on redo log files that are never filled up between your backups. When your data has no value, go ahead, otherwise switch to archive log mode.

Archives are created when a redo log group fills up. So, in your case you need to copy the online redo log files manually to the remote site for recovery.

How sure are you about the redo log files not being overwritten?

Be sensible and if this is production switch to archive log mode. Otherwise, promise not te make promises about being able to make point in time recoveries.

An other point: if your online redo log files are damaged, your database has a big problem and in your case you might loose a day worth of work. Is that OK? If not, reduce the size of the redo log files to a limit where it does make a switch every now and then. I am sure your company has an idea about how much time they can accept loosing transactions from. Many companies allow less than one hour transaction loss.