Checkpointing

Checkpointing is the primary technique that IMS uses to record information that can be used to restart an interrupted operation. Using the status information recorded during a checkpoint, IMS restores the contents of the message queues and database changes.

Checkpoints are an integral part of system shutdown and startup. Also, the amount of reprocessing, back from the point of system interruption and forward to a continuation point, is reduced when checkpointing is reasonably frequent. Some processing overhead is associated with checkpoint information, but this is an acceptable trade-off for the efficient restart of the system.

In an XRF complex, SNAPQ checkpoint records taken on the active IMS system are used to build control blocks on the alternate IMS system during the synchronization phase.

IMS internal checkpoints are scheduled to occur automatically at predetermined intervals. The interval is specified in terms of an increment to the number of system log records created. As the online IMS events are logged with individual log record types, a count is maintained. When the increment exceeds the specified value, checkpoint processing is invoked. IMS checkpoints can also be invoked explicitly by the master terminal operator and by application programs that have been authorized to issue the ⁄CHECKPOINT command.

Fast DB Recovery regions monitor checkpoint records on IMS systems, and uses them during database resource recovery.