Guidelines for templates and tape data sets
When you use the TEMPLATE utility to allocate tape data sets, use the STACK option to control tape processing.
STACK NO specifies traditional, single-file processing. The data set is written, and the tape is rewound and repositioned or even remounted. STACK YES specifies that successive files are to be written on a single logical tape without repositioning or remounting.
When you specify STACK YES, Db2 has the following behavior:
- Db2 stacks files only within a single utility invocation. When that utility ends, the stack is terminated, which means that the tape is rewound and unloaded. To allow more stacking, use the LISTDEF utility to define a list and then specify that list in another utility control statement. Using a LISTDEF list forces multiple objects to be processed under a single utility invocation.
- To preserve parallel processing, parallel tasks are written to
different tape volumes. The specific volume to which the data set
is written can vary, depending on the following factors:
- The number of output data sets that are being produced
- The number of parallel processes that are requested
- The number of tape units that are available to the job step
If you specify STACK YES, take the following actions as needed:
- To avoid issues with mounting a tape volume on the wrong drive, specify UNCNT 1.
- To prevent conflicts between parallel processes, use a single process to write a file to a stack. (Parallel processing can complicate stacking.)
- Ensure that only files of the same type are stacked on the same tape. For example, one tape might contain local primary image copies whereas another tape might contain remote primary image copies. The file types cannot be mixed.
- Ensure that the number of TEMPLATE utility control statements that specify STACK YES is no more than 85 for one invocation of the utility.
- For concurrent copies (copies that are made by the COPY utility with the CONCURRENT option)
- For inline image copies that are created by REORG TABLESPACE on a partition-by-growth base table space with one or more LOB columns
For data sets that might be redirected to disk by automatic class selection (ACS) rules
FL 508For COPYDDN or RECOVERYDDN data sets when running REORG to materialize MOVE TABLE operations
The
data sets and utilities for which the STACK YES option are supported
are listed in the following table. Yes
indicates that the specified
utility supports tape stacking for the specified data set. No
indicates
that the specified utility does not support tape stacking for the
specified data set. Ignored
indicates that the specified data
set does not apply to the specified utility.
ddname | CHECK DATA | CHECK INDEX or CHECK LOB | COPY | COPY- TOCOPY | LOAD | MERGE-COPY | REBUILD INDEX | REORG INDEX | REORG TABLE-SPACE | UNLOAD |
---|---|---|---|---|---|---|---|---|---|---|
SYSREC | Ignored | Ignored | Ignored | Ignored | No | Ignored | Ignored | Ignored | Yes | Yes |
SYSDISC | Ignored | Ignored | Ignored | Ignored | No | Ignored | Ignored | Ignored | Yes | Ignored |
SYSPUNCH | Ignored | Ignored | Ignored | Ignored | Ignored | Ignored | Ignored | Ignored | Yes | Yes |
SYSCOPY | Ignored | Ignored | Yes | Yes | No | Yes | Ignored | Ignored | Yes | Ignored |
SYSCOPY2 | Ignored | Ignored | Yes | Yes | No | Yes | Ignored | Ignored | Yes | Ignored |
SYSRCPY1 | Ignored | Ignored | Yes | Yes | No | Yes | Ignored | Ignored | Yes | Ignored |
SYSRCPY2 | Ignored | Ignored | Yes | Yes | No | Yes | Ignored | Ignored | Yes | Ignored |
SYSUT1 | No | No | Ignored | Ignored | No | Ignored | No | No | No | Ignored |
SORTOUT | No | Ignored | Ignored | Ignored | No | Ignored | Ignored | No | No | Ignored |
SYSMAP | Ignored | Ignored | Ignored | Ignored | No | Ignored | Ignored | Ignored | Ignored | Ignored |
SYSERR | No | Ignored | Ignored | Ignored | No | Ignored | Ignored | Ignored | Ignored | Ignored |
FILTERDDS | Ignored | Ignored | No | Ignored | Ignored | Ignored | Ignored | Ignored | Ignored | Ignored |