Duplex tape option

The duplex tape option provides an alternative to TAPECOPY processing for backup and migration cartridge-type tapes. Two tapes are created concurrently with one designated the original, the other the alternate. The intent is that the original tape be kept onsite, while the alternate can either be created in a remote tape library, or taken offsite.

The alternate tape must have the same tape geometry as the original. For example, if the original tape is a 3590-1 tape, the alternate must be also.

The duplex tape option also requires one additional tape drive per writing task. Adequate numbers of drives must be available for the tasking level used by the functions using duplexing during the time periods those functions run. This includes recycle.

With the SETSYS DUPLEX command you can specify that the duplex tape option be used for backup tapes only, migration tapes only, or both backup and migration tapes. If you use duplex creation for both migration and backup and you have been auto scheduling tape copying, you probably want to remove the auto scheduling. If you are duplexing either migration or backup tapes but not both, you can auto schedule tape copies based on the type indicated in the ARC0421I message.

In an SMS environment, the ACS routines can direct the alternate tape to the same or to a different tape library. Different tape libraries can exist at the same location, or a remote location. ESCON® technology defines how distant the remote location can be. In a non-SMS environment, the output restrictor is used for both the original and the alternate. ACS filtering can be set up to cause the desired tape allocations to occur.

Tapes created by the duplex tape option are compatible with tapes created by TAPECOPY.

DFSMShsm initial tape selection selects tapes for the duplex option using the following selection process:

A partial tape and its alternate are selected.
If no partial tapes with alternates are found, an empty ADDVOLed volume with a scratch volume as an alternate are selected.
If there are no empty ADDVOLed volumes, two scratch volumes are selected.

When problems occur that allow an original tape to be created, but prevent the alternate from being created by the duplex tape option, DFSMShsm automatically schedules internal TAPECOPY processing to ensure that valid alternate tapes exist. If the original tape is demounted before being filled it is marked as duplexed but unavailable for output until the alternate is successfully created. At either the EOV for the original volume or the end of the task, DFSMShsm schedules TAPECOPY processing to create an alternate tape. If that is unsuccessful, automatic scheduling of internal TAPECOPY processing occurs during secondary space management for migration volumes, and during automatic backup processing on the primary host for backup volumes.

As part of the automatic internal scheduling of TAPECOPY processing, a tape copy needed (TCN) record is written in the OCDS. When the automatic internal TAPECOPY completes successfully, the TCN record for the source volume (which is the original volume of the duplexed pair) is deleted, and the original tape volume is made available for reuse.

In a SETSYS PARTIALTAPE(REUSE) environment, both the duplexed original and its alternate can be reused (extended). If you have a non-duplexed original with an alternate created by TAPECOPY, only the original can be reused. The alternate is discarded.

The SETSYS DUPLEX command also allows you to specify that alternate and original migration tapes are to be marked full if an error occurs when writing to the alternate tape volume. To do so, enter the SETSYS DUPLEX(MIGRATION(Y command with the ERRORALTERNATE(MARKFULL) parameter specified:

SETSYS DUPLEX(MIGRATION(Y ERRALT(MARKFULL)))

The default for ERRORALTERNATE is CONTINUE. Here, if the alternate tape volume encounters an I/O error, the alternate is demounted and returned to scratch, but DFSMShsm continues writing to the original. When the writing is completed, DFSMShsm schedules a TAPECOPY request to create a new alternate tape.

A new flag has been added to the ARCTVEXT installation exit parameter list. When an alternate tape is being replaced and returned to scratch, and if the ARCTVEXT installation exit is active, the flag indicates if the action is the result of a user-issued TAPECOPY command. This information is provided so users are aware that an alternate tape is being returned to scratch, and that the original tape is not.

You can use the LIST and HLIST commands to show duplex tape status and alternate tape volume serial numbers.

Duplex tape does not support SETSYS TAPEUTILIZATION NOLIMIT set for the output unit being duplexed. This is because the alternate tape must match the original tape exactly, and with SETSYS TAPEUTILIZATION set to NOLIMIT this is impossible to guarantee, since either tape could be longer than the other. When TAPEUTILIZATION is set to NOLIMIT, the duplex tape option causes the tapes to reach a logical EOV when they are 97% full.

The duplex tape option does not support creating duplex copies of ML2 tapes during ARECOVER processing. Instead, one copy of the ML2 tape is recovered by ARECOVER processing, and a TCN record is written in the OCDS. The TCN record causes a TAPECOPY to be scheduled during the next occurrence of secondary space management processing. The TAPECOPY is not scheduled immediately because a TAPECOPY running during the recovery process might not be wanted.

The duplex tape option handles errors in the following way:

When an attempt is made to allocate and mount a partially full duplexed tape, but the associated alternate volume cannot be found, both the original and alternate are rejected and marked as unavailable for the duration of DFSMShsm processing.
When an attempt is made to allocate and mount an empty original tape and a new alternate scratch tape, but the scratch tape cannot be allocated or mounted, the duplex tape option continues to write the original volume. When the original volume (ADDVOLed or scratch) is demounted, DFSMShsm marks the volume as duplexed and unavailable for output, and schedules an internal TAPECOPY for that volume. DFSMShsm makes the original volume available after internal TAPECOPY processing successfully creates an alternate tape volume.
If the duplexed original encounters an I/O error, both the original and alternate tapes are marked full, demounted, and two different tapes are selected. The user data set related to the I/O error is restarted from the beginning on the different tapes.
If the alternate tape volume encounters an error, duplex tape processing depends on whether you specified the ERRORALTERNATE(MARKFULL) option on the SETSYS DUPLEX(MIGRATION) command. If so, both the original and alternate tapes are marked full if the alternate tape volume encounters an error. This option applies to migration tapes, for both initial migrations and subsequent RECYCLE processing. The option is not available for backup tapes. The ERRORALTERNATE(MARKFULL) option is recommended for installations that use higher capacity tapes and perform tape duplexing for disaster recovery purposes.
If you choose not to use this option, the default error handling is in effect. If the alternate tape volume encounters an error, the alternate is demounted and returned to scratch, but the original continues to be written. The original is marked as duplexed and unavailable for output, and an internal TAPECOPY is scheduled to automatically create an alternate tape. DFSMShsm makes the original volume available after internal TAPECOPY processing successfully creates an alternate tape volume.
When the original tape reaches its percent-full capacity, DFSMShsm must flush all the data to both tapes before performing a FEOV. If natural EOV is reached on either tape, DFSMShsm marks the tapes as FULL and restarts processing of the current data set from its beginning on two new tapes.