Command processing phases

To avoid using the most restrictive type of access for the duration of a command, processing for each command is divided into phases. This way, SMP/E can obtain resources using the proper access type at the start of each phase, and free them at the end of each phase. This minimizes the amount of time a resource is tied up for exclusive use.

These are the important things to remember about command processing phases:

Each command has at least two phases: initialization and termination. Most commands have one or more additional phases. For each command, the “Zone and Data Set Sharing Considerations” sub-topic of the topic in this information devoted to that command describes the various phases of that command, which resources are required, and the type of access needed for that resource.
During various phases of processing a command, SMP/E can use all three types of access for a given zone. For instance, during APPLY processing, the global zone is opened to obtain the name of the target zone CSI data set (read without enqueue), is then checked for which SYSMODs are eligible (read with shared enqueue), and is finally updated (update with exclusive enqueue).
The most important use of command phases is to manage the global zone. This is important because each command must at least access the global zone to obtain further processing information, and most of the commands also update the global zone at some time during processing. Without command phases, almost all background SMP/E jobs would be serialized on the CSI data set containing the global zone.