SnapVault for NetBackup with active/active configurations

Starting with Data ONTAP 7.2.1, it was possible to configure NetBackup storage units for volumes residing on systems that were part of an active/active configuration, but there was no failover support for SnapVault for NetBackup. Starting with Data ONTAP 7.3, failover for SnapVault for NetBackup functionality is supported.

To know how a takeover and giveback scenario works in terms of SnapVault for NetBackup, see the following example. Before going through the example, it is recommended that you review the general information about Data ONTAP active/active configurations in the Data ONTAP Active/Active Configuration Guide.

Takeover/giveback example for SnapVault for NetBackup: The following assumptions have been made:

A NetBackup disk storage unit, DSU_A, points to a volume residing on SystemA.
SystemA is part of an active/active configuration, and SystemB is its partner.

What happens if SystemA fails while a backup is in progress to DSU_A?: When SystemA fails, all connections from NetBackup to SystemA are terminated. SystemB detects that SystemA is down and initiates a takeover (that is, SystemB takes over SystemA). After the takeover is complete, NetBackup retries all requests that were terminated when SystemA went down. Automated retries of failed backups can be configured in NetBackup. (For more information on how to configure automatic retries, see the NetBackup documentation.)

Note: No configuration change is required in NetBackup in the event of a takeover, because SystemA can still be contacted on the same IP address even while it is taken over by SystemB. The takeover appears to NetBackup as a restart of SystemA.

What happens while SystemA continues to be down and SystemB is in takeover mode? When in takeover mode, SnapVault for NetBackup backup, restore, and all other operations initiated from NetBackup to the taken-over node (SystemA) are handled the same way as they are when not in takeover mode. Everything appears same from a NetBackup perspective, except for the following limitation that applies to SystemA while it is taken over by SystemB.

The number of concurrent connections are shared between SystemA and SystemB while in takeover mode. Performance might decrease for both SystemA and SystemB while in takeover mode, because the services for both SystemA and SystemB are now handled with the resources of a single system.

Backups sent to DSU_A during takeover mode are written to the same set of disks as before SystemA failed. It is only the physical entity controlling these disks that change on takeover. Thus, restores work regardless of whether the backup to be restored is taken prior to or after SystemA’s failure.

What happens when SystemA’s system comes online? When the system on which SystemA originally resided comes online, SystemB can initiate a giveback to SystemA. For information on how to initiate a giveback, see the Data ONTAP Active/Active Configuration Guide. The giveback of SystemA appears similiar to a node restart for NetBackup. Ongoing NetBackup connections are interrupted. After the giveback operation is complete, you must reinstantiate these connections from NetBackup.

Note: The connections to SystemB are not affected by the giveback operation.

Parent topic: About SnapVault for NetBackup