When a CICS region starts up after a previous shutdown, the CICS region performs recovery processing to return the region to the same state that it was in when it was last shut down. The CICS region automatically decides whether to perform a warm start (following a normal shutdown) or an emergency restart (following a system failure). You can force a cold start (in which the system log is reset during initialization), but data integrity problems can occur; for example, from in-doubt transactions that are not being recovered.
During a restart, the system data that was loaded into operating system memory during the previous run is used. For example, the resources that are in the autostart database are used (and not reloaded from the permanent database).
Recovering the state of recoverable resources after a system failure requires an external record of all the work that needs to be redone. For this purpose, CICS® is configured to take checkpoints periodically of the states of all recoverable resources. On restart, CICS reads the checkpoint to reestablish the states of the recoverable resources at the time the checkpoint was written, then processes all relevant information that is held for the region. You can control the frequency with which checkpoints are taken.