Maintaining a consistent image before restarting a Global Mirror or Metro Mirror relationship

When a consistent relationship is stopped, the relationship enters the consistent_stopped state. While in this state, I/O operations at the primary site continue to run. However, updates are not copied to the secondary site. When the relationship is restarted, the synchronization process for new data is started. This process does not preserve the original write ordering of the application. Unless extra steps are taken during this process, the relationship is in the inconsistent_copying state. The secondary volume for the relationship cannot be used until the copy process completes and the relationship returns to the consistent state.

To avoid this situation, you can configure a change volume for the secondary volume, or for each secondary volume if the relationship is part of a group. When a change volume is configured, the system automatically starts a FlashCopy to preserve a consistency copy of the secondary volume. The system also uses this image where required for disaster recovery.

Alternatively, you can manually start a FlashCopy operation for the secondary volume before you restart the relationship. Multiple-cycling Global Mirror relationships do not have this restriction because this cycling mode always maintains a consistent image while resynchronizing. While the relationship is in the copying state, the FlashCopy feature can provide a consistent copy of the data. If the relationship does not reach the synchronized state, you can use the FlashCopy target volume at the secondary site.

The SVCTools package, provides example scripts that demonstrate how to manage the FlashCopy process. See the copymanager script that is available in the SVCTools package. To use the developerWorks scripts, the automatic rebalancing feature and Easy Tier must be disabled by using the CLI.

You can download the SVCTools package from the following Web site: Scripting tools for SAN Volume Controller