Flash Copy for disaster recovery

The Flash Copy function is an LI REQ operation performed at the host that groups one or more clusters into a disaster recovery (DR) "family”. The DR family contains all the clusters in a DR site or sites and can be used to test DR preparedness.

Note: DR families should not be confused with Cluster Families.

To use Flash Copy for DR, configure “DR Families” via LI REQ at the host, then enable write protect or flash from the LI REQ command against all clusters in a DR family. Write protection is enabled across all clusters in a DR family, which permits them to move to the write protected state at the same time.

When the volume repository for all Clusters in a DR family are flash copied a snapshot (flashed instance) is created for all content in that repository. Changes that occur to volumes after the snapshot is created are updated in the live file system, while the original files remain unchanged and in the read-only state. This function is carried out among all Clusters in a DR family at the same time, which is referred to as "time zero". The snapshot process does not duplicate data, manages only two instances of each file. When a snapshot is created, changes to files are saved as a delta. Each subsequent host access references either the live version or the snapshot version, based on the selective write protect state. Host mounts to write-protected volumes access the snapshot, while volumes excluded from write protect access live data.

Restrictions:
  • Clusters in a DR family are assumed to remain operational in the event of a production outage.
  • Though a TS7700 Tape Attach Cluster can exist in a DR family, volumes in migrated state that belong to a TS7700 Tape Attach Cluster are not flash copied.
  • For a Tape Attach Cluster the flash copy will be performed only for the cache partitions. Volumes cannot be flash copied to physical tapes.
  • If no cluster has a valid copy at time zero within the DR family, the DR host mount fails.
  • Tape Volume Cache selection of snapshot data is limited only to those clusters in the DR family.
  • Snapshot data within a DR family is not replicated to peer DR family clusters. However, a mount to a write protected volume at the DR cluster creates a DR instance of ownership that is exchanged among peer clusters in a DR family. This allows DR host mounts and live production mounts to occur in parallel.
  • If a TS7700 Tape Attach Cluster exists in the DR family, its live content can be accessed only if no TS7700 Cluster in the DR family has a snapshot instance and the TS7700 Tape Attach Cluster is shown to have had a copy before time zero. You must invoke an LI REQ command to enable the use of the live copy before starting the mount operation. This step permits the microcode to access live content on the TS7700 Tape Attach Cluster in the DR family.
  • Volume removal is supported within flashed TS7700 Clusters, but the HSM component of the TS7700 must initiate the removal of the Flash Copy instances of that file. Delete-expired volumes are deleted in the flash snapshot only if that option is set to TRUE using the LI REQ command. Deleting data within the flash snapshot too early can cause the DR test to fail given the volumes are still viewed as private and accessible within the DR host-plex. Removal of a file from a snapshot is not supported; after the DR test completes, the volume flash is dissolved through a single LI REQ command.