First of all an excuse: these days I mainly use rman for backup and recovery tasks. I have used user managed backup and recovery in the past but it all has fade away….
Last Thursday and Friday I was at a client site to setup and test snapmanager for oracle (smo). With this tool you can use Netapp snapshots to backup Oracle databases and create “virtual” clone databases (a db clone without needing the disk space of the original db). Giving a full explanation of what netapp snapshots are and how smo works will lead too far for this post (I am working on a separate post about it), so just think about smo as a user managed backup.
Now this client wanted to extend the functionality of smo and wanted the snapshots (backups) to be replicated to a remote site. These replicated snapshots would then be used to create test/dev databases and for disaster recovery purposes.
One of my tasks was to write a disaster recovery procedure.
In short smo is performing the following steps when taking a full db backup:
- Place the entire database in backup mode.
- Take snapshots of all the netapp volumes containing database files. (think: take an os copy of all the data located on volumes containing data files).
- Take the database out of online backup mode.
- Issue an archive log current.
- Generate backup control files.
- Take a snapshot of the volumes containing the archive log files and the backup controlfile (think: take an os copy of all the data located on volumes containing archived redologs).
This backup data is then copied to a remote site.
