Improved data recording capability (IDRC) is a data compaction
feature of the 3480 subsystem, and part of the base model of the 3490
and 3490E cartridge controller. Data compaction in the cartridge control
unit eliminates the CPU overhead from software compression. IDRC-compacted
data is created using a technique called auto-blocking, which
ensures optimal physical blocking for cartridge media. The block size
is unaffected by this technique. A typical compaction ratio is 3:1;
however, inefficiently blocked tape data sets can achieve more dramatic
compaction.
IDRC helps to reduce the number of cartridges required to store
data, and reduces the elapsed time for batch jobs depending on cartridge
I/O. This makes IDRC effective for both single volume and multivolume
data sets. It optimizes the data exchange between
the controller and the device, increasing the number of devices that
can be used concurrently.
To implement IDRC with SMS data classes, perform the following
steps:
- Define data classes for your offsite tape data and very large
backup, active, and temporary tape data sets. Set COMPACTION to Y
for these data classes. Data classes for these tape data set categories
are TAPOSITE, TAPBKUP, TAPACTV, and TAPTEMP.
- Allow your data class ACS routine to assign these classes during
allocation. This method writes the data in IDRC-compacted format.
The same result occurs if you specify DCB=TRTCH=COMP on the DD statement.
Remember that the data class ACS routine is driven for both system-managed
and non-system-managed data sets.
Using data class lets the
system determine an optimal block size for the tape data set if you
do not specify one. z/OS DFSMS Using Data Sets describes
how the system determines block size for tape. Using system-determined
block size improves tape channel usage and buffer management.
- Define a data class, NONTMM, with COMPACT=N, so tape data sets
are directed to tape in a non-compacted form. Use the NONTMM data
class on your DD statements to tell your data class ACS routine that
the data sets should not be compacted or redirected to the DASD buffer.
These might be data sets shipped offsite to facilities without IDRC-capable
drives or those used by applications that call the READ BACKWARDS
command. This command is simulated for IDRC data sets; compacting
them severely degrades performance. Few data set types rely on the
READ BACKWARDS command. For example, the IMS™ log
is read backwards if it is processed during recovery. Do not compact
DFSORT work files.