HP XP Continuous Access Software - Manual (Page 99)

Continuous Access XP user guide for the XP1024/XP128 99
by means of multiplication can function with the Data fence-level setting. For example, Oracle multiplies
the Redo log file by itself (default = three times).
Figure 57 Relationship between log file and data file in PAIR status
Because most UNIX file systems (excluding JFS and VxFS) have no journal files, the P-VOL fence level
should be defined as Never. When a take-over by the S-VOL occurs, fsck is executed on the volume and
the file system is cleaned up, even if the S-VOL is undefined at the secondary host. Data lost depends on
how much differential data is contained in the P-VOL when the S-VOL is suspended. During operation,
error recovery should be performed when the suspended status (PSUE) is detected (when one error
occurs).
Transferring sense information between sites
When the MCU (or RCU for Continuous Access XP Async) suspends a Continuous Access XP pair due to
an error condition, the MCU/RCU sends sense information with unit check status to the appropriate hosts.
This sense information is used during disaster recovery to determine currency of the S-VOL, and must be
transferred to the remote site via the host failover software.
File and database recovery procedures
When a Continuous Access XP Synchronous pair is suspended, or when the MCU fails due to a disaster,
the S-VOL might contain in-process data. A data set could be open or transactions might not have
completed. Even if you use the S-VOL Data fence level for all Continuous Access XP Sync pairs, you must
establish file recovery procedures. These procedures should be the same as those used for recovering
volumes that become inaccessible due to control unit failure. These procedures are more important if the
S-VOL Status or Never fence-level settings are used.
Continuous Access XP Asynchronous does not provide procedures for detecting and retrieving lost
updates. To detect and recreate lost updates, check other current information (for example, database
journal log file active at the primary system when the disaster occurred). The Continuous Access XP Async
group consistency time (C/T) (MCU disk array service processor time) can be useful when performing this
detection and retrieval. Because this detection and retrieval process can take a while, your disaster
recovery scenario should be designed so detecting and retrieving lost updates are performed after the
application is started at the secondary system.
Prepare for file and database recovery using:
• Files for file recovery (such as database log files verified as current). To ensure currency of these files,
use the S-VOL Data fence level setting for Continuous Access XP pairs containing these important files.
• Sense information with local time-stamp transferred via host failover.
NOTE: Remote copy and disaster recovery procedures are inherently complex. Consult your HP account
team on sense-level settings and recovery procedures.
Remember that the Continuous Access XP Async time-stamp and C/T information indicates MCU service
processor (SVP) time, not host system time.
See ”
Pinned track recovery for Continuous Access XP volumes ” on page 103 for information about
recovering a pinned track on a Continuous Access XP volume.