Data Set Stacking

This chapter includes the following topics:

• Overview

• Before Implementation

• Troubleshooting

OverviewLink copied to clipboard

Stacking is the process by which Control-M/Tape places data sets with similar attributes (for example, retention) together on a volume. Data set stacking is a concept unique to Control-M/Tape. Therefore, it is recommended that you not implement this feature until Control-M/Tape is already running in Global Production mode (meaning, the old tape management system is no longer active at your site).

Two types of data set stacking can be performed by Control-M/Tape.

Table 11 Data Set Stacking Types

Through Data Set Stacking, Control-M/Tape can optimize usage of the storage space in your tape libraries.

Batch StackingLink copied to clipboard

The CTTSBD Control-M/Tape utility enables you to stack data sets already on tapes at your site. This utility can be used to free scratch tapes, override stacking limitations imposed by your real-time stacking definitions, and organize data sets that existed before your site’s tapes were managed by Control-M/Tape.

To immediately benefit from Control-M/Tape stacking capabilities, you can use the CTTSBD utility to stack existing data sets at your site before implementing Dynamic data set stacking (described below).

For more information about batch stacking, see the CTTSBD utility in the Control-M/Tape chapter of the INCONTROL for z/OS Utilities Guide.

Dynamic Data Set StackingLink copied to clipboard

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.

Control-M/Tape analyzes characteristics (for example, retention criteria and vault patterns) of each new data set, and searches for a volume with data sets with matching criteria and enough free space. The new data set is then added to the volume. This process is called dynamic data set stacking.

This feature is controlled by the DYNSTK Control-M/Tape installation parameter and the DO STACK statement that must appear in a Control-M/Tape rule for the data set.

For a general description of the data set stacking logic, see the description of stacking in the organization and administration chapter of the Control-M/Tape User Guide. The following discussion assumes an understanding of the stacking algorithms and operation of Control-M/Tape.

General Considerations for Dynamic Data Set StackingLink copied to clipboard

Before activating the Dynamic Data Set Stacking facility, verify that the parameters relevant to the Dynamic Data Set Stacking facility are set during installation.

In general, Control-M/Tape Dynamic Data Set Stacking facility selects a volume for stacking only if the following conditions are met:

• The DYNSTK parameter is set toY in the CTTPARM member and a DOSTACK=Y statement is specified in a matching rule.

• The job names and data set names of the creating and created jobs have a matching entry in the Stacking Database. If you want to change this criteria (for example, consider all data sets with a specific prefix as the same data set for stacking purposes), you can use the CTTX002 Dynamic Data Set Stacking facility user exit to override the default operation of Control-M/Tape.

• The average size megabytes of the specified data set is read from the appropriate entry in the Stacking Database. If this information cannot be found, stacking cannot continue. To avoid this problem, specify a default size for data sets through the STKDEFSZ parameter in the CTTPARM member.

  The value of the STKDEFSZ installation parameter can optionally be overridden for a data set or group of data sets through a DO STKDEFSZ statement in a Control-M/Tape rule.

• An appropriate volume is found in the Media Database. This volume must be eligible for stacking, have sufficient free space, and belong to the same pool as the data set to be stacked. Depending on the specification for the STKMODE Control-M/Tape parameter, the volume used for stacking may also need to have the same vaulting pattern as the data set to be stacked.

  The search for a suitable volume can be controlled by Exit 10 (Find Stackable Volume) that controls the stacking algorithm. For more information about Exit 10, refer to the CTTX010 member in the IOA SAMPEXIT library.

  The search can be limited to a specific number of volumes through the STKSRCHL parameter in the CTTPARM member or through a DO STKSRCHL statement in a Control-M/Tape rule.

• The data set to be stacked is allocated in a DDstatement in the JCL of a job. Control-M/Tape stacks data sets that are allocated dynamically (using SVC99) only if the STKALCD parameter was set to Y in the CTTPARM member.

  For a more complete list of conditions, considerations and parameters, see the description of stacking in the organization and administration chapter of the Control-M/Tape User Guide.

• Non-scratch (Active) volumes converted to Control-M/Tape from another Tape Management System are considered not eligible for stacking. Stacking is only performed for volumes that were moved from scratch status to non-scratch status (meaning, the first data set was created on the scratch volume) while Control-M/Tape was operational.

  Active volumes converted from the other tape management system can be stacked through the CTTSBD batch stacking utility (described above). For more information see the description of the CTTSBD utility in the Control-M/Tape chapter of the INCONTROL for z/OS Utilities Guide.

  The remainder of this section describes how to implement the Dynamic Data Set Stacking facility at your site..

Before ImplementationLink copied to clipboard

As with any automated process to be implemented system-wide, the key to efficient and trouble free stacking is proper preparation. Procedures must be established and standards should be set.

Below is a description of the basic preparatory steps that should be performed before dynamic data set stacking is implemented at your site.

Review Working ProceduresLink copied to clipboard

Determine the standards for data organization at your site.

Typical questions that should be answered are:

• What kind of retention is used?

• What standards are used to determine vault patterns?

• What naming conventions are used? For example, which prefixes are assigned to data sets created by the various applications in use at your site.

• Are pools used at your site? If so, which pools are assigned to the various applications and/or departments in use at your site?

Identify Data Sets That Should Not be StackedLink copied to clipboard

Some data sets need to be insulated from other data sets in your system. These data sets must be identified and excluded from implementation of the Dynamic Data Set Stacking facility. Types of data sets that should not be stacked include:

• Data sets created by software products that expect their data sets to be stored on separate volumes (for example, CA-View/Express files).

• Data sets that need to be sent to other data centers or systems that are not recognized by local vaulting procedures (as described above).

• Data accessed by other systems (MVS or non-MVS) that may be updated without Control-M/Tape being notified. For example, an updated data set may not be recorded in the MVS Catalog by the application or system that is creating the data set.

Select Test Environment for Stacking ImplementationLink copied to clipboard

It is recommended that you first implement the Dynamic Data Set Stacking facility for a limited number of data sets in a specific environment.

Testing a single job with a single data set is usually the first step in activating the Dynamic Data Set Stacking facility.

After this test, you can choose to implement stacking for additional data sets according to application, data set name, and so on.

Technical PreparationLink copied to clipboard

Review Installation ParametersLink copied to clipboard

Review the Control-M/Tape installation parameters relevant to the Dynamic Data Set Stacking facility and verify that appropriate values are specified. If you change any of these parameters except for DYNSTK and STKUNIT, reload it into the real-time environment using the following command:

S CTTINIT,PARM=‘RELOAD,TBLT=PARM’

If you change the value of the DYNSTK parameter or the STKUNIT parameter, the new values take effect only after Control-M/Tape is stopped and then restarted.

Table 12 Installation Parameters

STKUNIT=(unit1,unit2,.....)

CTTSTKLink copied to clipboard

Verify that the CTTSTK utility is not commented out in the New Day procedure (CTTDAY).

If the CTTSTK utility was commented out, BMC recommends that you run this utility on the Media Database (it is normally run on the Trace file) to collect the statistics needed for stacking implementation.

Research Your SystemLink copied to clipboard

• Analyze the Media Database using utility CTTRPT. A wide variety of reports can be generated through this utility. Some examples of how this utility can be used to extract information for stacking implementation are listed below.

  • List under-utilized volumes (through the CTTRUTIL sample report). The volumes in this report are sorted by used capacity to help identify the best candidate volumes for stacking implementation.

  • Sort the inventory by data set name. Usually this results in sorting the data sets according to application (due to naming conventions).

    Sample CTTRPT reports are located in the IOA SAMPLE library. For more information about the CTTRPT utility, see the Control-M/Tape chapter of the INCONTROL for z/OS Utilities Guide.

• Identify candidate data sets for stacking implementation. Selection of data sets should depend on:

  • the type of application that generated the data sets

  • which data sets belong to an application that frequently encounters a shortage of scratched volumes

  • the size of relevant data sets

  • which data sets are assigned to each pool (if pools are in use at the site)

• If you are working with STKMODE set toR orA, data sets for which retention is not a specific date or number of days (for example, retention type CATALOG, CYCLES, or LAST ACCESS) are stacked according to their average life span. The Dynamic Data Set Stacking facility calculates the expected expiration date of each such data set (from information in the Stacking Database) and stacks them according to the requested search algorithm.

  If you prefer that data sets with non-specific retention be handled in a different way, generate a list of these data sets through the CTTRPT report utility, and define Control-M/Tape rules that include the DOSTKMODE=S statement for these data sets.

• Identify potential conflicts (meaning, data sets that should not be stacked together). The CTTSCA Control-M/Tape utility can be used to automatically detect certain types of conflicts and to generate rules that prevent those conflicts from occurring. For more information, see the Control-M/Tape chapter of the INCONTROL for z/OS Utilities Guide.

• Identify groups of data sets that should reside together. These data sets should be identified as stacking groups through DO STKGROUP statements in Control-M/Tape rules.

ImplementationLink copied to clipboard

If Control-M/Tape was brought up without activating the Dynamic Data Set Stacking facility, you have to stop Control-M/Tape and restart it in order activate this facility.

Determine which data sets, jobs, and tape pools should be stacked.

• Verify that the appropriate rules have been defined, coded, and activated (ordered) for these data sets, jobs, or tape pools.

• The DOSTACK=Y statement must be included in rules that refer to the data sets that are to be stacked. Rules modified to support stacking must be reloaded into the Control-M/Tape Real-time environment using the following command:

S CTTINIT,PARM=‘RELOAD,TBLT=RULE’

• Use the CTTCRSS ISPF online utility to ensure that the stacking rules you are invoked for the appropriate jobs and data sets. (TSO%CTTCRSS, or from the IOA Online Utility menu, specify Option6, followed by OptionT1.)

• Make sure that all stacking rules are in production mode (that is, MODE is set to PROD in the rule definition).

• Use DOSTKRULE statements in the rules to prevent possible data set conflicts (see the CTTSCA utility in the Control-M/Tape Utilities chapter of the INCONTROL for z/OS Utilities Guide).

• Use DOSTKGROUP statements in the rules to define logical groups of data sets. Each data set is stacked only with other data sets that have been assigned the same group name.

For more information about specific DO statements in Control-M/Tape rules, see the rule parameters chapter of the Control-M/Tape User Guide.

For more information about conditions necessary for stacking implementation, see the description of stacking in the organization and administration chapter of the Control-M/Tape User Guide.

Stacking RulesLink copied to clipboard

Dynamic data set stacking can be controlled by a variety of different DO statements in Control-M/Tape rules. These DO statements enable specification of stacking criteria that affect only the data set or data sets processed by a specific rule. The DO statements that can be specified are shown in Table 13.

Table 13 Dynamic Stacking DO Statements

For more information about these DO statements, see the rule parameters chapter of the Control-M/Tape User Guide.

TroubleshootingLink copied to clipboard

If the Dynamic Data Set Stacking facility is not working as expected, use the following steps to find and fix the problem:

Identify Problem Data Set and run Rule SimulationLink copied to clipboard

Identify the data set, job, and so on for which stacking is not working as expected.

Use ISPF online utility CTTCRSS (rule simulation) to determine which rules are triggered for the data set or job. Verify that each of the appropriate rules is in production mode (that is, MODE is set to PROD), and that the DO STACK=Y statement is included.

Check JCLLink copied to clipboard

Verify that the data set names, job names, and so on specified in the JCL are consistent with the names specified in the relevant Control-M/Tape rules. Ensure that DISP is set to (NEW,CATLG) for the data set. Alternatively, ensure DISP is set to (NEW,KEEP), and STKKEEP is set to r, in the CTTPARM member.

Check MVS Jobname MonitoringLink copied to clipboard

The Control-M/Tape subsystem receives control when message IEF403I ("jobname STARTED") is issued. Verify that MVS monitoring of job names is turned on (that is, that this message is not suppressed).

If necessary, command MONITOR JOBNAMES can be used to turn on MVS jobname monitoring. This command should also be included in the CONSOLxx member in the SYS1.PARMLIB library (or any equivalent) to ensure that the command is automatically issued at each IPL.

Check Stacking Statistics CollectionLink copied to clipboard

Verify that the job name in the JCL is the same as the job name for which statistics were collected by the CTTSTK utility. (This may not be the case if data sets with the same name were created by two different jobs.)

Make sure that the CTTSTK utility is included as a step in the New Day procedure (CTTDAY). This utility reviews the Control-M/Tape Trace file and collects the statistics needed for operation of the Dynamic Data Set Stacking facility.

This information can also be obtained by running the CTTSTK utility, using the Control-M/Tape Trace file as input.

The CTTSTKR utility can also be used to list the contents of the Stacking Database. For more information on the CTTSTKR utility, see the Control-M/Tape chapter of the INCONTROL for z/OS Utilities Guide.

Check Exit 2Link copied to clipboard

Control-M/Tape Exit 2 (CTTX002) can be used to modify the name of a job that creates a data set, the data set name and so on, so that certain data sets are considered as successive generations of the same data set in the Stacking Database. Sample Exit CTTX002R demonstrates how this exit can manipulate dsnames and jobnames to allow specification of stacking attributes according to dsnames and jobnames with shorter prefixes. For example, this exit can be used to group data sets with the same jobname and different suffixes (for example, to ignore timestamps in data set names).

Incorrect coding of Exit 2 may cause dynamic data set stacking problems.

For more information about this exit, see the CTTX002R and CTTX002 members in the IOA SAMPEXIT library.

Check Unit NameLink copied to clipboard

Verify that the unit specified in the JCL has been specified through the STKUNIT Control-M/Tape installation parameter in the CTTPARM member.

Check for a //NOSTACK DD StatementLink copied to clipboard

Make sure that a //NOSTACK DD statement is not included in the JCL.

Determine why the Data set was not StackedLink copied to clipboard

If, after making the above checks, a data set was still not stacked as expected, review the reasons why the expected action was rejected.

Control-M/Tape checks each prospective stacking action in three stages:

1. JCL (DDstatements)

2. data set considerations

3. volume considerations

4. dynamic allocation (SVC 99) considerations

Each of these stages can result in rejection of the action. Each rejection is assigned a reason code. (The rejection reasons for each stage are described later in this chapter.

Normally only rejection reasons due to data set considerations are visible to the user. Each time a data set is rejected for stacking, the CTT356W message is issued. This message is followed by a message that describes why the data set was not stacked. (Data set rejection reasons are issued by the CTTSTH Control-M/Tape module.)

If a data set was not stacked but no CTT356W messages were issued, stacking may have been rejected for all volumes in the specified pool. To display rejection reasons for volumes, specify a CTTSTKRL DD statement (for example, CTTSTKRL DD DUMMY) in any step of the job that creates the data set to be stacked. This DD statement causes a rejection reason to be issued for each volume that is rejected during the search for a volume on which to stack the data set.

If no data set or volume rejection reason is found to explain a failed stacking attempt, stacking may have been rejected due to a job statement or dynamic allocation consideration. Review the job statement rejection reasons below or the dynamic allocation rejections to understand the cause of the problem.

S CTTINIT,PARM='TL=324,TJ=job-name'

S CTTINIT,PARM='TL=0'

Certain specifications in the DD statements of a job may make it impossible to stack data sets accessed or created by the job. The table shown below lists each rejection reason code and a description of the problem it indicates.

Table 14 Job Statement Rejection Reason Code

Data Set Rejection ReasonsLink copied to clipboard

As mentioned above, data set rejection reasons are listed in the job log following CTT356W messages.

Table 15 Data Set Rejection Reason Code

Volume Rejection ReasonsLink copied to clipboard

The following rejection reasons indicate that a specific volume was rejected during the search for a volume on which to stack the current data set. These rejection reasons do not necessarily mean that the data set was not stacked. Many volumes may be considered and rejected during the search for a volume on which to stack a data set.

Table 16 Volume Rejection Reason Code

Dynamic Allocation ConsiderationsLink copied to clipboard

Certain specifications in the dynamic allocation (SVC 99) parameters may make it impossible to stack data sets. The table shown below lists each rejection reason code and a description of the problem it indicates.

Table 17 Dynamic Allocation Consideration Code

Real-Time Data Set Stacking Under JES3Link copied to clipboard

JES3 normally issues all tape mount requests in immediate mode at the beginning of a job. However, it is possible to implement Real-time data set stacking under JES3.

In order to implement Real-time data set stacking mount requests under JES3, the mount requests must be intercepted and modified. Use JES3 sample Exit IATUX06 to perform the necessary modifications.

When a job starts, Exit IATUX06 analyzes all tape requests in the job. The exit changes mount requests to DEFERRED mode and issues a message if:

• The data set that is to be created is a candidate for stacking.

• The mount request is in immediate mode (meaning, DEFER is not specified).

When the data set is created, Control-M/Tape searches for an appropriate volume and issues a mount message.

A data set is a candidate for stacking if all of the following are true:

• The request is for a scratch tape.

• Control-M/Tape is active and real-time data set stacking is enabled (meaning, the DYNSTK installation parameter is set to Y).

• A corresponding Control-M/Tape rule has been defined and is active. This rule must include a DOSTACK=Y statement.

ImplementationLink copied to clipboard

Use the following steps to implement real-time data set stacking under JES3:

1. Edit the ASMIAT06 member in the CTT JCL library and submit the job. When the job is submitted, the ASMIAT06 member compiles and links JES3 Exit IATUX06. Exit IATUX06 is called by JES3 when each DDstatement is processed.

2. Copy the new IATUX06 module to the JES3 exits library (SYS1.JES3LIB).

3. To activate the new exit, stop and restart JES3 using Hot Start (an IPL is not required).