Dynamic Dataset Stacking

When a scratch volume is requested for a new data set, Control-M/Tape can automatically direct the data set to an active volume that contains one or more files with attributes similar to the new data set.

This feature is controlled by both the DYNSTK installation parameter, which much appear in the CTTPARM installation parameter, and the DO STACK statement, which must appear in the Control-M/Tape rule of the data set.

At the start of a job, Control-M/Tape searches the JCL of the job for a request for a new tape data set to be created on a scratch volume. If such a request is found, Control-M/Tape verifies the following:

• Control-M/Tape is operating in Global PHASED / PROD Mode.

  When Control-M/Tape is operating in TEST mode, Dynamic Dataset Stacking is controlled by the STKTEST parameter in the CTTPARM member.

  Note: The CTTPARM member can be refreshed in memory by starting the CTTINIT procedure with the expression parm=`RELOAD,TBLT=PARM’.

• The matching rule does not specify the expression RETENTION=EDM or MODE=TEST.
• A matching record for the data set is found in the Stacking Database.
• The name of the data set to be found can be changed by Exit 2.

This check can be overridden when parameter STKDEFSZ is specified in member CTTPARM or in a rule.

• The data set is marked as stackable in the Stacking Database.
• The pool name specified in the rule points to an active (defined) pool.

Depending on the results of the check, either of the following occurs:

• If any of these checks fails, message CTT356W is issued followed by a message that identifies the cause of the failure, and stacking is not performed for the data set.
• If all checks succeed, the Media Database is scanned for a suitable volume. The maximum number of volumes that can be scanned is specified in parameter STKSRCHL in CTTPARM. Either of the following two results is possible:
• If the search limit is reached without finding a suitable volume, stacking is not performed, message CTT356W is issued followed by a message that identifies the reason that stacking could not be performed, and a SCRATCH request (mount request for a scratch volume) is issued.
• If a suitable volume is found, the job’s parameters are adjusted to direct the data set onto the selected volume. Instead of issuing a SCRATCH request, the mount message requests the selected volume.

The search for a suitable volume process calls Exit 10 (Find Stackable Volume Exit). This exit can be used to control the search algorithm. For more information regarding Exit 10, refer to "Control-M/Tape Exits" in the Exits chapter of the INCONTROL for z/OS Administrator Guide, and to member DOCTX010 in the IOA DOC library.

Each subsequent search for a suitable volume begins from the last volume scanned in the preceding search.

The space required for the new data set is estimated based on the sizes of previous versions of the data set. Previous versions must have the same data set name and must be created by the same job name. This information is kept in the Stacking Database.

For every newly created data set, Control-M/Tape writes a record containing data set attributes (such as block size and block count) in the Trace file. During the New Day procedure, a utility (CTTSTK) reads the Trace file and updates the information in the Stacking Database.

If a data set is recreated, processed with the expression DISP=MOD, or processed in a step with DD statement //NOSTACK, an entry is created in the Stacking Database marking this data set as non-stackable. The volume on which this data set resides is also marked as non-stackable (in the Media Database).

Control-M/Tape maintains the used space (megabytes used) and free space (megabytes remaining) of each volume. Whenever a volume becomes scratch, it is assigned the maximum size, according to its media. This size is specified in the MEDIA statement in member CTTPARM.

Besides having enough space and belonging to the appropriate pool, the selected volume can be forced to meet one or both requirements controlled by parameter STKMODE in member CTTPARM:

• The volume can be forced to have the same vault pattern as the data set (or not have any vault pattern if the data set is not intended for vaulting).
• The last data set on the volume can be forced to have a retention period that is equal to or greater than that of the new data set.

User Exit CTTX002 can be used to change default processing (for example, it can eliminate the job name factor so that any data set with the same name is considered a previous version of the new data set).