Disasters occur in many ways. In a sudden disaster, all volumes
may be in normal full duplex mode when you begin recovery procedures.
In such a case it is easy to determine how current the data is.
A disaster can also occur over a period of time. This rolling disaster
is the most common situation, and necessitates that you check each
volume to determine how current the data is. SYSLOG messages written
by the ERP give the times when specific volumes are suspended.
It is important to understand how a rolling disaster can affect
your primary system. This knowledge enables you to avoid a potential
data integrity exposure when using PPRC in a disaster recovery situation.
A disaster or series of disasters can result in intermittent failures
to the primary system. In this situation, you might be unable to determine
which volumes are recoverable because of one of the following scenarios:
- A disaster could leave many or all of your secondary volumes in
a suspended state. Without the IEA49xx message
to indicate when these volumes became suspended, each volume would
require extensive analysis to determine how current the data is.
- If the attempt to update the secondary volume fails due to a link
failure, PPRC might be unable to mark the secondary volume as suspended.
- The operating systems process all error notification actions asynchronously.
Positive identification and recording of an event, such as broken
PPRC ESCON® links, happen after
the event itself.
During a disaster recovery operation, the secondary volume would appear to be recoverable. There would, however,
be no way to be certain that all of the data is current. The way to
minimize (or even avoid) this potential data integrity exposure is
by applying the disaster recovery preparations that are described
in Considering the PPRC solution and Preparing for PPRC error recovery. Failover/failback for a Global Mirror session describes
how to judge whether recovery data is current.