When production is interrupted by an outage at the primary site,
Failover/Failback eliminates the need to transmit all the data back
from the recovery site for a full volume copy when the primary site
comes back online and is ready to resume production. Following an
outage, the primary volumes are often in the same state they were
in when the outage occurred. In such a case, the only difference
between the primary volumes and the secondary volumes are the updates
made to the secondary volumes since the outage occurred.
When an outage occurs, Failover processing causes the secondary
volumes to become suspended primary volumes. The Failover PPRC secondary
must be in DUPLEX,
SUSPENDED, or PENDING-XD state. Any subsequent
changes made to those volumes are recorded by the subsystem. When
the primary site is capable of resuming production, Failback processing
transmits only those changes back to the primary site. The Failover
process makes the following assumptions, but operational capabilities
may vary and these conditions are not required for the process to
function:
- All paths between the PPRC primary and secondary subsystems are
inoperable
- Paths cannot be used from the secondary subsystem to the primary
subsystem
- The PPRC secondary has updated data consistent with other PPRC
secondary volumes
Note: - If the device used for failover processing cannot also be used
for failback (such as when the storage controller is irreparably damaged
and must be replaced), then failback cannot take place and a full
volume copy must be executed.
- If the secondary is in a soft fence state, failback processing
clears the soft fence state.