Recovery with data sharing

In a data-sharing environment, you must extend IMS recovery procedures by protecting surviving subsystems from database records that contain incomplete changes left by a failing subsystem. IMS recovers a shared database by restoring the database from the latest image copy.

You can use the merge function provided with the IMS Database Change Accumulation utility (DFSUCUM0) to merge the log data sets from the subsystems that updated the database.

Apply the records from the utility (in the output data sets) for a forward recovery. If only one subsystem in a data-sharing environment updated the database, IMS does not need to use the sequence numbers and the Database Change Accumulation utility, because it has only one set of log data sets to manage.

IMS data sharing also introduces other challenges. Two IMS Internal Resource Lock Managers (IRLMs) might lose communication with each other because:

• An z/OS® cross-system coupling facility (XCF) might fail.
• An IRLM might fail, leaving its IMS subsystems running.
• A z/OS system might fail, bringing down the IRLM, VTAM®, DBRC, and IMS subsystems running under it.

In a data-sharing environment, IMS must ensure data integrity after failure of one or more components of data sharing.

The IRLM and DBRC work together with IMS to protect shared databases before and after a failure leaves incomplete changes in these databases. This protection remains after a failure until failed subsystems complete backout of incomplete changes.