z/OS DFSMS Implementing System-Managed Storage
Previous topic | Next topic | Contents | Contact z/OS | Library | PDF


Managing Non-GDG Batch Data Sets

z/OS DFSMS Implementing System-Managed Storage
SC23-6849-00

Because production cycles vary greatly, consider their characteristics carefully when you plan for automatic management by DFSMShsm. The MONTHMIG management class handles batch application data sets generated during a monthly cycle. Table 1 lists its attributes.

Few characteristics of batch data sets can be generalized. Consider these variances before you automate space and availability management with DFSMShsm:
  • Batch data sets might be inactive for a long time, but when needed, must be available immediately. This type of data set should not be migrated to tape.
  • Batch data set sizes might vary greatly, based on the production cycle. These data sets should be assigned a management class with Partial Release=NO to inhibit DFSMShsm from releasing space.
  • Batch data sets might only be needed for a short period. These data sets might consist of reports or error listings, and be eligible for early deletion.

You should supplement the basic set of management classes with others that reflect the data set requirements for specialized applications.

If you have a large, complex batch workload and want to use SMS performance services without analyzing the management requirements of the data, you can migrate these data sets to system management and assign a management class that does not cause any automatic space or availability management actions to occur. Always assign a management class, because if one is not assigned, the default management class attributes or DFSMShsm defaults are used. Once you have migrated your batch data, you can design the management classes and the management class ACS routines to accommodate batch data. You can assign the management classes for batch by executing DFSMSdss's CONVERTV, using the REDETERMINE option for volumes containing batch data.

The sample management class routine detects generation data sets using the DSTYPE read/only variable. It then assigns a management class, depending on whether the generation data set is production or backup. A data set naming standard that uses the second-level qualifier to convey whether a data set is a backup or production GDS is assumed in the sample ACS routine. Figure 1 shows the management class ACS routine logic for batch data sets. As with other data types, a large percentage of batch data is assigned the STANDARD management class.

Go to the previous page Go to the next page




Copyright IBM Corporation 1990, 2014