User's Manual

Consider the following scenarios where A is the Primary System, C the Asynchronous Periodic
Backup System, and B the Synchronous Backup System, and SRM/SRA is configured between HP
3PAR StoreServ Storage systems A and C.
• All the links are up:
SRM initiates the data transfer from A to C through HP 3PAR SRA. After the sync is complete,
SRA stops the RC groups between A—B and also between A—C. Then SRA initiates failover
at C so that C becomes the Failover System and takes the role of the Primary System.
• A—C link is down:
HP 3PAR SRA initiates the data transfer from B to C.
After the sync is complete, SRA stops the RC groups between A—B. Then SRA initiates failover
at C so that C becomes the Failover System and takes the role of the Primary System. Also
see the “Limitations of SLD configuration” (page 38) section.
• A—B link is down:
When the A—C link is up:
SRM initiates the data transfer from A to C through HP 3PAR SRA. Once the sync is complete,
SRA stops the RC groups between A—C. SRA then initiates a failover at C so that C becomes
the Failover System and takes the role of the Primary System.
When A—C link is down:
Depending on the situation, either B or C will have the most current data since the synchronous
link between A—B is down. In such a scenario, as per the design, HP 3PAR SRA does not
initiate any data transfer from B to C. With the available data, C becomes the failover system
and takes the role of the primary system.
CAUTION: If C does not contain the most current data then any data replicated to B that is
not replicated to C is discarded.
• B—C link is down:
When A—C link is up:
SRM initiates the data transfer from A to C through HP 3PAR SRA. Once the sync is complete,
SRA stops the RC groups between A—B and also between A—C. SRA then initiates failover
at C so that C becomes the failover system and takes the role of the primary system.
When the A—C link is down:
HP 3PAR SRA will not initiate the data transfer between B—C, because the B—C standby link
is not available for data transfer. SRA stops the RC groups between A—B. SRA then initiates
failover at C. With the available data, C becomes the Failover System and takes the role of
the Primary System.
CAUTION: If C does not contain the most current data and any data replicated to B that is
not yet replicated to the C is discarded.
Reprotect
Perform the reprotect operation to configure protection in the reverse direction (from Site B to Site
A), as a preparation for failback to the original state. The SRA operations are similar to the planned
migrations, and deletes the earlier snapshots that were created during failover from Site A to Site
B.
For SLD configurations, the reprotect operation requires that the remote copy links between the
new primary and both the targets to be up. If the array C is the new primary system and if the links
24 Recovery and failback