System checkpoints

IMS automatically takes periodic system checkpoints, so if it is necessary to restart IMS, it can begin at the last checkpoint (or an earlier one if the work done before the checkpoint was not complete).

The interval between system checkpoints is measured by how many log records IMS writes, so the checkpoints are a measure of activity rather than elapsed time. You select the interval between system checkpoints when you install IMS.

As with most choices involving preparation for recovery, the impact of taking frequent checkpoints must be weighed against the advantages of faster recovery. Because IMS does not suspend work to take a checkpoint, decreased transaction speed is not a significant impact factor.

Recommendation: Do not suppress system checkpoints to improve system performance because emergency restart must always go back at least two system checkpoints (or back to a prior restart).

The IMS MTO can also request an additional system checkpoint at any time by issuing a /CHECKPOINT command.

Commands, such as /DBR, /DBD, /STA, and /STO, will be rejected and message DFS140I is issued if a checkpoint is in progress when those commands are entered.

IMS takes a system checkpoint when any of the following occur:

IMS starts
A regular interval has elapsed
The MTO enters a command (/DBRECOVERY, UPDATE DB STOP(ACCESS) OPTION(FEOV), /CHECKPOINT, or /SWITCH OLDS CHECKPOINT)
IMS shuts down

If IMS is enabled with resource definition data sets or the IMSRSC repository and AUTOEXPORT=AUTO, RDDS, or REPO is specified in the DFSDFxxx PROCLIB member, IMS will automatically export to the RDDS or the repository any changes in resource definitions for MODBLKS resources since the last automatic export or the EXPORT DEFN command at the end of each normal checkpoint processing.