- SYSTEM_DEFAULTS. The lowest priority specification. Any constraint(s) specified on any matching level selector will override corresponding values specified here. The values specified here are used only if none of the matching level selectors specify a corresponding constraint. For example, the SYSTEM_DEFAULT specified value for MAXBUFSP is used only if no matching level selector specifies MAXBUFSP.
- SMS classes (SCLASS, MCLASS, DCLASS), from low to high precedence.
This is the least specific level that you can specify (except for the system defaults). Many clusters may be defined in each class. Each cluster defined in a particular class can be accessed by several jobs, each of which is composed of several job steps. Each of these job step programs performs a different type of processing against the cluster. Therefore, limits specified at the SMS class level are not specific, but general guidelines for dynamic buffering to observe, unless overridden by a more specific level selector match.
- CLUSTER. This level is only slightly more specific than SMS storage, management, or data class. Many jobs can access this cluster. Each job can have several steps or programs that may access this cluster. All of these programs and jobs perform different types of processing against this cluster. Therefore, limits specified at the cluster level are only slightly more specific than SMS classes.
- JOBNAME. This is more specific than cluster. A job can be composed of several steps, and some or all of these steps can process different clusters. Each program is probably related to the others in that they are all probably part of the same application system (such as, payroll) and may or may not perform similar processing. For example, step one of a job might randomly update a master file. Step two of the same job might sequentially copy that same master file to tape.
- PROGRAM name. More specific than job name. A particular program may access more than one VSAM file, with different types of processing for each file. For example, one of the clusters may be a master file which is processed randomly, and another cluster a journal file that is processed sequentially.
Both the program name on the EXEC card and the program name on the CDE are checked.
- DDNAME combined with program name. This is the most specific level available. A particular program (for instance, the PAYROLL program) always performs the same type of processing against the cluster that is allocated by a particular DD card (such as, the PAYMSTR DD). This is true regardless of which cluster is named on that DD card, or which job is running this program. For example, a PAYROLL program may be used by a service bureau to create paychecks for many customers. Each customer has a separate master file, so this program is run many times, each time with a different cluster specified on the PAYMSTR DD card.
However, these clusters are all similar in that each has the same record layout (if not, the payroll program would certainly encounter problems), record length(s), and key specification, and so forth. The processing is approximately the same. The payroll program might read sequentially through the file and update year-to-date values in the records as it creates the paychecks. This is true for any PAYMSTR DD cluster.
