Using GO SAVE Option 21 in a Geographic Mirrored Environment

To obtain a valid system save via GO SAVE option21 in a Geographic Mirrored environment (Same concept for Metro Mirror or Global Mirror), the following procedures can be used:

*This document assumes the following - Start with an Active Geographic Mirrored environment (IASP is Available on Production node, clustering on both nodes is active, device CRG is active, Geographic Mirroring Actively synching).

First Procedure: GO SAVE option 21 on Production Node
1. Suspend the replication of Geographic Mirroring (See note below for Detach consideration)
2. End the device CRG (This will ensure no accidental failover occurs)
3. End clustering on the Production Node
4. Perform GO SAVE option 21
5. After the save has been performed and QCTL has been restarted, start clustering back up
*NOTE - If there is built in automation of starting cluster back up, there should be nothing further to do
6. Start clustering on the Production Node
7. Start the device CRG
8. Resume the Geographic Mirroring

Note: In the above procedure, the PRODUCTION copy is being saved rather than the mirror copy. The users cannot access the mirror copy of the IASP. All access to the IASP must be through the production copy. This means the production copy is not available for use during the save process and requires an actual outage window.
Additionally, in step 1 where the Suspend is being suggested, as a protective measure a Detach could be used instead.  Doing a Detach would allow for an accessible copy of the IASP on the mirror copy in the event that some failure occurs on the source during the backup and renders the primary copy no longer accessible.  Manual steps could be used to failover to the mirror in that case.

An alternative save procedure that allows the production copy to remain usable but requires a partial resync at the end involves review of the Second Procedure below.

Steps 3 through 6 below might be more drastic than necessary. The user could do a *FULL save of the IASP without actually restricting the system if desired. All caveats about keeping users off the mirror copy IASP also apply. This allows a save of the IASP without ever ending the production copy. The users of the system will not experience any downtime during this process. However, the IASP is running exposed during the save and resync process, meaning that there is no High Availability during this procedure.

Second Procedure: GO SAVE option 21 on Detached Mirror Copy Node
1. Quiesce IASP on Production Node first using CHGASPACT command
2. Detach geographic mirror copy
3. Vary on the IASP on the detached mirror copy
4. End the device CRG
5. End the mirror copy cluster node.
6. Perform GO SAVE option 21
7. When save is complete, prepare to re-establish geomirror synchronization
8. Vary off the detached mirror copy IASP.
9. Start the mirror copy cluster node
10. Start the device CRG
11. Reattach geographic mirroring ( A partial resync will be initiated)

Note: When using a Detach with tracking option, the amount of resynchronization is minimized, but it is expected that there will be "some" resync. The information above in the first procedure is simply a method to avoid resynchronization altogether; however, it requires downtime.

When external storage is involved, FlashCopy is another consideration at either production or mirror sites.  Taking a FlashCopy can be done while suspending geographic mirror only for a small moment. This would reduce the amount of time to resync and allow the the environment to become highly available, all while a backup runs on the FlashCopy target node.