Introduction
It is quite common to have different views on backups between DBAs and Netbackup admins. DBAs want to be sure that backups are being performed correctly and often prefer to make backups to disk with their own tools and after that Netbackup will do a backup to another repository to ensure backups lifecycle. However, this double copy has several problems:
- There is no control over the validity of the backup because there is no single process that controls everything
- A double copy affects both backup window and system performance
- In order to restore it is also necessary to make a double step
Copilot for Oracle is a solution that aims to satisfy both DBAs and Netbackup admins since its operation is similar to what they are currently doing but controlled by Netbackup:
- Netbackup Appliance export a file share by NFS
- The Oracle server mounts the resource by NFS
- A backup of the Oracle database is done on this share
- Everything is coordinated by Netbackup that will handle that backup later to another device (MSDP, Cloud, Tape, etc ...)
Thus:
- The dba has a copy to disk and can use Oracle tools to restore
- It avoids punishing the production disk system since the disk copy is on an appliance outside the production disk array and at lower cost
- No double restoration is necessary
- It is possible to make accelerated backups because it integrates with the Oracle Incremental Merge funcionality. Thus, once it has made a full backup, subsequent copies only consist of changed blocks that are consolidated to generate a complete new copy. In addition, taking advantage of the Appliance file system, snapshots with full copies in the Appliance are created and managed
Configuration
Copilot needs a supported Appliance (currently 5230 or 5330) with version 2.7.1 or higher. We need to configure a share on the Appliance. We can do it from the "clish" in the "Manage-> Storage" menu with the "Create Share" option:
This option will create and configure the permissions for the resource to be shared by NFS. Then you must configure this share on the Oracle server resource:
You may also need to change /etc/fstab (on the Oracle server) installation in order to remount the share after rebooting the server.
We will also need 2 slps: one slp for snapshots, with a first operation "Snapshot" and then a "Backup From Snapshot":
A second slp with a single backup operation:
We can then create the Oracle policy. We will choolse the SLP we created that has the snapshot (in this case slp_stream):
After that the we will choose the instances we want to backup:
Under "Backup Selections" we choose "Whole Database - Datafile Copy Share" and type the share:
And finally for the schedule we will select the SLP with the snapshot operation:
After executing the policy, we can see that the share has a copy of the Oracle databases:
And from the point of view of Oracle it is also possible to see copies:
We will see several jobs in the "Activity Monitor" corresponding to the backup (and subsequent duplication), and after the first copy we can get faster thanks to accelerated copy:
We can also refer to "job detail" optimization with the "accelerator":
We see therefore that this feature brings numerous improvements, not only technical, but also to allow better relationship between DBAs and backup administrators.
References
https://www.veritas.com/support/en_US/article.000100516
Source document in Spanish in my Netbackup blog:
http://netbackupguru.es/netbackup-copilot-para-oracle/
